You are viewing a plain text version of this content. The canonical link for it is here.
Posted to dev@skywalking.apache.org by 13240156934 <13...@163.com> on 2020/05/20 09:53:17 UTC
sw-bug in tomcat-7.x-8.x-plugins
hi god wu.sheng:
First of all, thank you for your reading.
I have a question, and it makes us rollback the skywalking deployment in our prod env.And I need you help.
The question is ,there is lot's of url in our spring-controller,and some is dynamic parameter, such as "http://localhost:8080/hello/user/{userID} ".
When the qps is growing with lot's of different userID,the sw-collector’s Memory will grow geometrically.Finally when the Old generation of JVM is Exhausted,the process will keep fgc forever. It's not an accident.
I had blocked for 2 weeks, and today, I found it may because of an plugin:"tomcat-7.x-8.x-plugins", when the TomcatInvokeInteceptor.beforeMethod creates span,
sw will set operationName by request.getRequestURI() ,in general,that should transfer to a variable such as "{userID}",but it not work.
I need your help, and look forward to your suggestion
Yours respectfully
2020.05.20
Re:Re: Re: Re: Re: sw-bug in tomcat-7.x-8.x-plugins
Posted by 13240156934 <13...@163.com>.
yes sir,I will follow this rule, it make me feel at home~
At 2020-05-21 15:19:15, "Sheng Wu" <wu...@gmail.com> wrote:
>Welcome, and when you have issues, please continue the discussion in the
>mail list.
>
>Sheng Wu 吴晟
>Twitter, wusheng1108
>
>
>13240156934 <13...@163.com> 于2020年5月21日周四 下午3:15写道：
>
>> Thankyou for you reply, I will keep follow on skywalking, and hope to do
>> my share for the skywalking.
>>
>>
>>
>>
>>
>>
>>
>> At 2020-05-21 14:40:59, "Sheng Wu" <wu...@gmail.com> wrote:
>> >Inline.
>> >
>> >13240156934 <13...@163.com> 于2020年5月21日周四 下午12:25写道：
>> >
>> >>
>> >>
>> >>
>> >> Oh god,I got the answer what I wanted.
>> >> I have an suggest,that the automatically override rule should consider
>> >> this situation,that if there is one or more than one aop spans between
>> >> tomcat-plugins and springmvc span,it should automatically override also.
>> >>
>> >
>> >This is very hard to implement.
>> >
>> >
>> >> Finally ,I have two question, the one is when does the 8.0 version
>> planned
>> >> to go online.
>> >
>> >
>> >The current plan is the end of this month. We will send the notifications
>> >of the release test and release vote in this mail list.
>> >
>> >
>> >> The other one is how can I know the detail that how the automatically
>> >> override works.
>> >>
>> >
>> >I will suggest you take a look `StackBasedTracingSpan` class. You should
>> be
>> >easy to understand why I said it is hard to implement, and how we did that
>> >today.
>> >
>> >Sheng Wu 吴晟
>> >Twitter, wusheng1108
>> >
>> >
>> >> Thanks a lot for your help,Thank you ,Thank you.
>> >>
>> >>
>> >>
>> >>
>> >>
>> >>
>> >>
>> >>
>> >>
>> >>
>> >>
>> >>
>> >>
>> >>
>> >> At 2020-05-21 11:17:57, "Sheng Wu" <wu...@gmail.com> wrote:
>> >> >Yes, that would break the automatically override.
>> >> >I think the better way to erase this issue, is removing the tomcat
>> plugin,
>> >> >if you really need `com.test.rest.aop.ApiTestNetAspect.around` span,
>> >> >because I don't know what is in that span.
>> >> >
>> >> >Sheng Wu 吴晟
>> >> >Twitter, wusheng1108
>> >> >
>> >> >
>> >> >13240156934 <13...@163.com> 于2020年5月21日周四 上午11:05写道：
>> >> >
>> >> >>
>> >> >>
>> >> >>
>> >> >> All rignt,the picture may like this:
>> >> >> /v3/market/200626/candles Tomcat
>> >> >> com.test.rest.aop.ApiTestNetAspect.around
>> >> >> /v3/market/{trument_id}/candles SpringMVC
>> >> >>
>> >> >> This up 3 lines are the simple trace, you will see the there is a aop
>> >> >> between two entryspan,and the first span's url is not dynamic
>> parameter.
>> >> >>
>> >> >>
>> >> >>
>> >> >>
>> >> >>
>> >> >>
>> >> >>
>> >> >>
>> >> >>
>> >> >>
>> >> >>
>> >> >> At 2020-05-21 10:55:33, "Sheng Wu" <wu...@gmail.com>
>> wrote:
>> >> >> >You can't send the attachment on the mail list, provide a link to
>> your
>> >> >> >image, please.
>> >> >> >
>> >> >> >> The second,the gc problem shouldn't happen in my prod env,because
>> my
>> >> url
>> >> >> >by dynamic parameter maybe only 10000,and costs about 10g mem is
>> >> accepted.
>> >> >> >> So I wonder,if I still use sw-v7.0 and open the cache and register
>> >> >> >mechanism, and I solve the dynamic parameter problem, Is it my
>> problem
>> >> >> that
>> >> >> >will not appear again?
>> >> >> >
>> >> >> >Once you have limited URI, and the memory cost is acceptable, then
>> it
>> >> is
>> >> >> >fine. Btw, there is no such thing `open the cache and register
>> >> mechanism`.
>> >> >> >It exists there for all releases before 8.0
>> >> >> >
>> >> >> >
>> >> >> >
>> >> >> >Sheng Wu 吴晟
>> >> >> >Twitter, wusheng1108
>> >> >> >
>> >> >> >
>> >> >> >13240156934 <13...@163.com> 于2020年5月21日周四 上午10:39写道：
>> >> >> >
>> >> >> >>
>> >> >> >> hi god wu
>> >> >> >>
>> >> >> >> I have understood what you mean.It seems that's not the bug in
>> >> >> >> tomcat-7.x-8.x-plugin,it may be that I have use aop around my
>> >> >> >> controller,such as this picture.
>> >> >> >>
>> >> >> >> The second,the gc problem shouldn't happen in my prod env,because
>> my
>> >> url
>> >> >> >> by dynamic parameter maybe only 10000,and costs about 10g mem is
>> >> >> accepted.
>> >> >> >> So I wonder,if I still use sw-v7.0 and open the cache and register
>> >> >> >> mechanism, and I solve the dynamic parameter problem, Is it my
>> >> problem
>> >> >> >> that will not appear again?
>> >> >> >>
>> >> >> >>
>> >> >> >>
>> >> >> >>
>> >> >> >>
>> >> >> >>
>> >> >> >> At 2020-05-20 19:18:11, "Sheng Wu" <wu...@apache.org> wrote:
>> >> >> >> >I am not sure why you say this is Tomcat plugin bug. If you are
>> >> using
>> >> >> >> >spring mvc and plugin of spring mvc, the uri with parameter
>> should
>> >> be
>> >> >> >> >detected and overrided the original tomcat url automatically.
>> >> >> >> >And the GC issue is another thing, all cache and register
>> mechanism
>> >> >> have
>> >> >> >> >been removed. No risk about that.
>> >> >> >> >
>> >> >> >> >Sheng
>> >> >> >> >
>> >> >> >> >13240156934 <13...@163.com>于2020年5月20日 周三下午5:53写道：
>> >> >> >> >
>> >> >> >> >> hi god wu.sheng:
>> >> >> >> >> First of all, thank you for your reading.
>> >> >> >> >> I have a question, and it makes us rollback the skywalking
>> >> >> deployment in
>> >> >> >> >> our prod env.And I need you help.
>> >> >> >> >> The question is ,there is lot's of url in our
>> >> spring-controller,and
>> >> >> some
>> >> >> >> >> is dynamic parameter, such as "
>> >> >> http://localhost:8080/hello/user/{userID}
>> >> >> >> >> ".
>> >> >> >> >> When the qps is growing with lot's of different userID,the
>> >> >> sw-collector’s
>> >> >> >> >> Memory will grow geometrically.Finally when the Old generation
>> of
>> >> >> JVM is
>> >> >> >> >> Exhausted,the process will keep fgc forever. It's not an
>> >> accident.
>> >> >> >> >> I had blocked for 2 weeks, and today, I found it may because
>> of an
>> >> >> >> >> plugin:"tomcat-7.x-8.x-plugins", when the
>> >> >> >> >> TomcatInvokeInteceptor.beforeMethod creates span,
>> >> >> >> >> sw will set operationName by request.getRequestURI() ,in
>> >> >> general,that
>> >> >> >> >> should transfer to a variable such as "{userID}",but it not
>> work.
>> >> >> >> >> I need your help, and look forward to your suggestion
>> >> >> >> >>
>> >> >> >> >>
>> >> >> >> >> Yours respectfully
>> >> >> >> >> 2020.05.20
>> >> >> >> >
>> >> >> >> >--
>> >> >> >> >Sheng Wu 吴晟
>> >> >> >> >
>> >> >> >> >Apache SkyWalking
>> >> >> >> >Apache Incubator
>> >> >> >> >Apache ShardingSphere, ECharts, DolphinScheduler podlings
>> >> >> >> >Zipkin
>> >> >> >> >Twitter, wusheng1108
>> >> >> >>
>> >> >> >>
>> >> >> >>
>> >> >> >>
>> >> >> >>
>> >> >> >>
>> >> >>
>> >>
>>
Re: Re: Re: Re: sw-bug in tomcat-7.x-8.x-plugins
Posted by Sheng Wu <wu...@gmail.com>.
Welcome, and when you have issues, please continue the discussion in the
mail list.
Sheng Wu 吴晟
Twitter, wusheng1108
13240156934 <13...@163.com> 于2020年5月21日周四 下午3:15写道：
> Thankyou for you reply, I will keep follow on skywalking, and hope to do
> my share for the skywalking.
>
>
>
>
>
>
>
> At 2020-05-21 14:40:59, "Sheng Wu" <wu...@gmail.com> wrote:
> >Inline.
> >
> >13240156934 <13...@163.com> 于2020年5月21日周四 下午12:25写道：
> >
> >>
> >>
> >>
> >> Oh god,I got the answer what I wanted.
> >> I have an suggest,that the automatically override rule should consider
> >> this situation,that if there is one or more than one aop spans between
> >> tomcat-plugins and springmvc span,it should automatically override also.
> >>
> >
> >This is very hard to implement.
> >
> >
> >> Finally ,I have two question, the one is when does the 8.0 version
> planned
> >> to go online.
> >
> >
> >The current plan is the end of this month. We will send the notifications
> >of the release test and release vote in this mail list.
> >
> >
> >> The other one is how can I know the detail that how the automatically
> >> override works.
> >>
> >
> >I will suggest you take a look `StackBasedTracingSpan` class. You should
> be
> >easy to understand why I said it is hard to implement, and how we did that
> >today.
> >
> >Sheng Wu 吴晟
> >Twitter, wusheng1108
> >
> >
> >> Thanks a lot for your help,Thank you ,Thank you.
> >>
> >>
> >>
> >>
> >>
> >>
> >>
> >>
> >>
> >>
> >>
> >>
> >>
> >>
> >> At 2020-05-21 11:17:57, "Sheng Wu" <wu...@gmail.com> wrote:
> >> >Yes, that would break the automatically override.
> >> >I think the better way to erase this issue, is removing the tomcat
> plugin,
> >> >if you really need `com.test.rest.aop.ApiTestNetAspect.around` span,
> >> >because I don't know what is in that span.
> >> >
> >> >Sheng Wu 吴晟
> >> >Twitter, wusheng1108
> >> >
> >> >
> >> >13240156934 <13...@163.com> 于2020年5月21日周四 上午11:05写道：
> >> >
> >> >>
> >> >>
> >> >>
> >> >> All rignt,the picture may like this:
> >> >> /v3/market/200626/candles Tomcat
> >> >> com.test.rest.aop.ApiTestNetAspect.around
> >> >> /v3/market/{trument_id}/candles SpringMVC
> >> >>
> >> >> This up 3 lines are the simple trace, you will see the there is a aop
> >> >> between two entryspan,and the first span's url is not dynamic
> parameter.
> >> >>
> >> >>
> >> >>
> >> >>
> >> >>
> >> >>
> >> >>
> >> >>
> >> >>
> >> >>
> >> >>
> >> >> At 2020-05-21 10:55:33, "Sheng Wu" <wu...@gmail.com>
> wrote:
> >> >> >You can't send the attachment on the mail list, provide a link to
> your
> >> >> >image, please.
> >> >> >
> >> >> >> The second,the gc problem shouldn't happen in my prod env,because
> my
> >> url
> >> >> >by dynamic parameter maybe only 10000,and costs about 10g mem is
> >> accepted.
> >> >> >> So I wonder,if I still use sw-v7.0 and open the cache and register
> >> >> >mechanism, and I solve the dynamic parameter problem, Is it my
> problem
> >> >> that
> >> >> >will not appear again?
> >> >> >
> >> >> >Once you have limited URI, and the memory cost is acceptable, then
> it
> >> is
> >> >> >fine. Btw, there is no such thing `open the cache and register
> >> mechanism`.
> >> >> >It exists there for all releases before 8.0
> >> >> >
> >> >> >
> >> >> >
> >> >> >Sheng Wu 吴晟
> >> >> >Twitter, wusheng1108
> >> >> >
> >> >> >
> >> >> >13240156934 <13...@163.com> 于2020年5月21日周四 上午10:39写道：
> >> >> >
> >> >> >>
> >> >> >> hi god wu
> >> >> >>
> >> >> >> I have understood what you mean.It seems that's not the bug in
> >> >> >> tomcat-7.x-8.x-plugin,it may be that I have use aop around my
> >> >> >> controller,such as this picture.
> >> >> >>
> >> >> >> The second,the gc problem shouldn't happen in my prod env,because
> my
> >> url
> >> >> >> by dynamic parameter maybe only 10000,and costs about 10g mem is
> >> >> accepted.
> >> >> >> So I wonder,if I still use sw-v7.0 and open the cache and register
> >> >> >> mechanism, and I solve the dynamic parameter problem, Is it my
> >> problem
> >> >> >> that will not appear again?
> >> >> >>
> >> >> >>
> >> >> >>
> >> >> >>
> >> >> >>
> >> >> >>
> >> >> >> At 2020-05-20 19:18:11, "Sheng Wu" <wu...@apache.org> wrote:
> >> >> >> >I am not sure why you say this is Tomcat plugin bug. If you are
> >> using
> >> >> >> >spring mvc and plugin of spring mvc, the uri with parameter
> should
> >> be
> >> >> >> >detected and overrided the original tomcat url automatically.
> >> >> >> >And the GC issue is another thing, all cache and register
> mechanism
> >> >> have
> >> >> >> >been removed. No risk about that.
> >> >> >> >
> >> >> >> >Sheng
> >> >> >> >
> >> >> >> >13240156934 <13...@163.com>于2020年5月20日 周三下午5:53写道：
> >> >> >> >
> >> >> >> >> hi god wu.sheng:
> >> >> >> >> First of all, thank you for your reading.
> >> >> >> >> I have a question, and it makes us rollback the skywalking
> >> >> deployment in
> >> >> >> >> our prod env.And I need you help.
> >> >> >> >> The question is ,there is lot's of url in our
> >> spring-controller,and
> >> >> some
> >> >> >> >> is dynamic parameter, such as "
> >> >> http://localhost:8080/hello/user/{userID}
> >> >> >> >> ".
> >> >> >> >> When the qps is growing with lot's of different userID,the
> >> >> sw-collector’s
> >> >> >> >> Memory will grow geometrically.Finally when the Old generation
> of
> >> >> JVM is
> >> >> >> >> Exhausted,the process will keep fgc forever. It's not an
> >> accident.
> >> >> >> >> I had blocked for 2 weeks, and today, I found it may because
> of an
> >> >> >> >> plugin:"tomcat-7.x-8.x-plugins", when the
> >> >> >> >> TomcatInvokeInteceptor.beforeMethod creates span,
> >> >> >> >> sw will set operationName by request.getRequestURI() ,in
> >> >> general,that
> >> >> >> >> should transfer to a variable such as "{userID}",but it not
> work.
> >> >> >> >> I need your help, and look forward to your suggestion
> >> >> >> >>
> >> >> >> >>
> >> >> >> >> Yours respectfully
> >> >> >> >> 2020.05.20
> >> >> >> >
> >> >> >> >--
> >> >> >> >Sheng Wu 吴晟
> >> >> >> >
> >> >> >> >Apache SkyWalking
> >> >> >> >Apache Incubator
> >> >> >> >Apache ShardingSphere, ECharts, DolphinScheduler podlings
> >> >> >> >Zipkin
> >> >> >> >Twitter, wusheng1108
> >> >> >>
> >> >> >>
> >> >> >>
> >> >> >>
> >> >> >>
> >> >> >>
> >> >>
> >>
>
Re:Re: Re: Re: sw-bug in tomcat-7.x-8.x-plugins
Posted by 13240156934 <13...@163.com>.
Thankyou for you reply, I will keep follow on skywalking, and hope to do my share for the skywalking.
At 2020-05-21 14:40:59, "Sheng Wu" <wu...@gmail.com> wrote:
>Inline.
>
>13240156934 <13...@163.com> 于2020年5月21日周四 下午12:25写道：
>
>>
>>
>>
>> Oh god,I got the answer what I wanted.
>> I have an suggest,that the automatically override rule should consider
>> this situation,that if there is one or more than one aop spans between
>> tomcat-plugins and springmvc span,it should automatically override also.
>>
>
>This is very hard to implement.
>
>
>> Finally ,I have two question, the one is when does the 8.0 version planned
>> to go online.
>
>
>The current plan is the end of this month. We will send the notifications
>of the release test and release vote in this mail list.
>
>
>> The other one is how can I know the detail that how the automatically
>> override works.
>>
>
>I will suggest you take a look `StackBasedTracingSpan` class. You should be
>easy to understand why I said it is hard to implement, and how we did that
>today.
>
>Sheng Wu 吴晟
>Twitter, wusheng1108
>
>
>> Thanks a lot for your help,Thank you ,Thank you.
>>
>>
>>
>>
>>
>>
>>
>>
>>
>>
>>
>>
>>
>>
>> At 2020-05-21 11:17:57, "Sheng Wu" <wu...@gmail.com> wrote:
>> >Yes, that would break the automatically override.
>> >I think the better way to erase this issue, is removing the tomcat plugin,
>> >if you really need `com.test.rest.aop.ApiTestNetAspect.around` span,
>> >because I don't know what is in that span.
>> >
>> >Sheng Wu 吴晟
>> >Twitter, wusheng1108
>> >
>> >
>> >13240156934 <13...@163.com> 于2020年5月21日周四 上午11:05写道：
>> >
>> >>
>> >>
>> >>
>> >> All rignt,the picture may like this:
>> >> /v3/market/200626/candles Tomcat
>> >> com.test.rest.aop.ApiTestNetAspect.around
>> >> /v3/market/{trument_id}/candles SpringMVC
>> >>
>> >> This up 3 lines are the simple trace, you will see the there is a aop
>> >> between two entryspan,and the first span's url is not dynamic parameter.
>> >>
>> >>
>> >>
>> >>
>> >>
>> >>
>> >>
>> >>
>> >>
>> >>
>> >>
>> >> At 2020-05-21 10:55:33, "Sheng Wu" <wu...@gmail.com> wrote:
>> >> >You can't send the attachment on the mail list, provide a link to your
>> >> >image, please.
>> >> >
>> >> >> The second,the gc problem shouldn't happen in my prod env,because my
>> url
>> >> >by dynamic parameter maybe only 10000,and costs about 10g mem is
>> accepted.
>> >> >> So I wonder,if I still use sw-v7.0 and open the cache and register
>> >> >mechanism, and I solve the dynamic parameter problem, Is it my problem
>> >> that
>> >> >will not appear again?
>> >> >
>> >> >Once you have limited URI, and the memory cost is acceptable, then it
>> is
>> >> >fine. Btw, there is no such thing `open the cache and register
>> mechanism`.
>> >> >It exists there for all releases before 8.0
>> >> >
>> >> >
>> >> >
>> >> >Sheng Wu 吴晟
>> >> >Twitter, wusheng1108
>> >> >
>> >> >
>> >> >13240156934 <13...@163.com> 于2020年5月21日周四 上午10:39写道：
>> >> >
>> >> >>
>> >> >> hi god wu
>> >> >>
>> >> >> I have understood what you mean.It seems that's not the bug in
>> >> >> tomcat-7.x-8.x-plugin,it may be that I have use aop around my
>> >> >> controller,such as this picture.
>> >> >>
>> >> >> The second,the gc problem shouldn't happen in my prod env,because my
>> url
>> >> >> by dynamic parameter maybe only 10000,and costs about 10g mem is
>> >> accepted.
>> >> >> So I wonder,if I still use sw-v7.0 and open the cache and register
>> >> >> mechanism, and I solve the dynamic parameter problem, Is it my
>> problem
>> >> >> that will not appear again?
>> >> >>
>> >> >>
>> >> >>
>> >> >>
>> >> >>
>> >> >>
>> >> >> At 2020-05-20 19:18:11, "Sheng Wu" <wu...@apache.org> wrote:
>> >> >> >I am not sure why you say this is Tomcat plugin bug. If you are
>> using
>> >> >> >spring mvc and plugin of spring mvc, the uri with parameter should
>> be
>> >> >> >detected and overrided the original tomcat url automatically.
>> >> >> >And the GC issue is another thing, all cache and register mechanism
>> >> have
>> >> >> >been removed. No risk about that.
>> >> >> >
>> >> >> >Sheng
>> >> >> >
>> >> >> >13240156934 <13...@163.com>于2020年5月20日 周三下午5:53写道：
>> >> >> >
>> >> >> >> hi god wu.sheng:
>> >> >> >> First of all, thank you for your reading.
>> >> >> >> I have a question, and it makes us rollback the skywalking
>> >> deployment in
>> >> >> >> our prod env.And I need you help.
>> >> >> >> The question is ,there is lot's of url in our
>> spring-controller,and
>> >> some
>> >> >> >> is dynamic parameter, such as "
>> >> http://localhost:8080/hello/user/{userID}
>> >> >> >> ".
>> >> >> >> When the qps is growing with lot's of different userID,the
>> >> sw-collector’s
>> >> >> >> Memory will grow geometrically.Finally when the Old generation of
>> >> JVM is
>> >> >> >> Exhausted,the process will keep fgc forever. It's not an
>> accident.
>> >> >> >> I had blocked for 2 weeks, and today, I found it may because of an
>> >> >> >> plugin:"tomcat-7.x-8.x-plugins", when the
>> >> >> >> TomcatInvokeInteceptor.beforeMethod creates span,
>> >> >> >> sw will set operationName by request.getRequestURI() ,in
>> >> general,that
>> >> >> >> should transfer to a variable such as "{userID}",but it not work.
>> >> >> >> I need your help, and look forward to your suggestion
>> >> >> >>
>> >> >> >>
>> >> >> >> Yours respectfully
>> >> >> >> 2020.05.20
>> >> >> >
>> >> >> >--
>> >> >> >Sheng Wu 吴晟
>> >> >> >
>> >> >> >Apache SkyWalking
>> >> >> >Apache Incubator
>> >> >> >Apache ShardingSphere, ECharts, DolphinScheduler podlings
>> >> >> >Zipkin
>> >> >> >Twitter, wusheng1108
>> >> >>
>> >> >>
>> >> >>
>> >> >>
>> >> >>
>> >> >>
>> >>
>>
Re: Re: Re: sw-bug in tomcat-7.x-8.x-plugins
Posted by Sheng Wu <wu...@gmail.com>.
Inline.
13240156934 <13...@163.com> 于2020年5月21日周四 下午12:25写道：
>
>
>
> Oh god,I got the answer what I wanted.
> I have an suggest,that the automatically override rule should consider
> this situation,that if there is one or more than one aop spans between
> tomcat-plugins and springmvc span,it should automatically override also.
>
This is very hard to implement.
> Finally ,I have two question, the one is when does the 8.0 version planned
> to go online.
The current plan is the end of this month. We will send the notifications
of the release test and release vote in this mail list.
> The other one is how can I know the detail that how the automatically
> override works.
>
I will suggest you take a look `StackBasedTracingSpan` class. You should be
easy to understand why I said it is hard to implement, and how we did that
today.
Sheng Wu 吴晟
Twitter, wusheng1108
> Thanks a lot for your help,Thank you ,Thank you.
>
>
>
>
>
>
>
>
>
>
>
>
>
>
> At 2020-05-21 11:17:57, "Sheng Wu" <wu...@gmail.com> wrote:
> >Yes, that would break the automatically override.
> >I think the better way to erase this issue, is removing the tomcat plugin,
> >if you really need `com.test.rest.aop.ApiTestNetAspect.around` span,
> >because I don't know what is in that span.
> >
> >Sheng Wu 吴晟
> >Twitter, wusheng1108
> >
> >
> >13240156934 <13...@163.com> 于2020年5月21日周四 上午11:05写道：
> >
> >>
> >>
> >>
> >> All rignt,the picture may like this:
> >> /v3/market/200626/candles Tomcat
> >> com.test.rest.aop.ApiTestNetAspect.around
> >> /v3/market/{trument_id}/candles SpringMVC
> >>
> >> This up 3 lines are the simple trace, you will see the there is a aop
> >> between two entryspan,and the first span's url is not dynamic parameter.
> >>
> >>
> >>
> >>
> >>
> >>
> >>
> >>
> >>
> >>
> >>
> >> At 2020-05-21 10:55:33, "Sheng Wu" <wu...@gmail.com> wrote:
> >> >You can't send the attachment on the mail list, provide a link to your
> >> >image, please.
> >> >
> >> >> The second,the gc problem shouldn't happen in my prod env,because my
> url
> >> >by dynamic parameter maybe only 10000,and costs about 10g mem is
> accepted.
> >> >> So I wonder,if I still use sw-v7.0 and open the cache and register
> >> >mechanism, and I solve the dynamic parameter problem, Is it my problem
> >> that
> >> >will not appear again?
> >> >
> >> >Once you have limited URI, and the memory cost is acceptable, then it
> is
> >> >fine. Btw, there is no such thing `open the cache and register
> mechanism`.
> >> >It exists there for all releases before 8.0
> >> >
> >> >
> >> >
> >> >Sheng Wu 吴晟
> >> >Twitter, wusheng1108
> >> >
> >> >
> >> >13240156934 <13...@163.com> 于2020年5月21日周四 上午10:39写道：
> >> >
> >> >>
> >> >> hi god wu
> >> >>
> >> >> I have understood what you mean.It seems that's not the bug in
> >> >> tomcat-7.x-8.x-plugin,it may be that I have use aop around my
> >> >> controller,such as this picture.
> >> >>
> >> >> The second,the gc problem shouldn't happen in my prod env,because my
> url
> >> >> by dynamic parameter maybe only 10000,and costs about 10g mem is
> >> accepted.
> >> >> So I wonder,if I still use sw-v7.0 and open the cache and register
> >> >> mechanism, and I solve the dynamic parameter problem, Is it my
> problem
> >> >> that will not appear again?
> >> >>
> >> >>
> >> >>
> >> >>
> >> >>
> >> >>
> >> >> At 2020-05-20 19:18:11, "Sheng Wu" <wu...@apache.org> wrote:
> >> >> >I am not sure why you say this is Tomcat plugin bug. If you are
> using
> >> >> >spring mvc and plugin of spring mvc, the uri with parameter should
> be
> >> >> >detected and overrided the original tomcat url automatically.
> >> >> >And the GC issue is another thing, all cache and register mechanism
> >> have
> >> >> >been removed. No risk about that.
> >> >> >
> >> >> >Sheng
> >> >> >
> >> >> >13240156934 <13...@163.com>于2020年5月20日 周三下午5:53写道：
> >> >> >
> >> >> >> hi god wu.sheng:
> >> >> >> First of all, thank you for your reading.
> >> >> >> I have a question, and it makes us rollback the skywalking
> >> deployment in
> >> >> >> our prod env.And I need you help.
> >> >> >> The question is ,there is lot's of url in our
> spring-controller,and
> >> some
> >> >> >> is dynamic parameter, such as "
> >> http://localhost:8080/hello/user/{userID}
> >> >> >> ".
> >> >> >> When the qps is growing with lot's of different userID,the
> >> sw-collector’s
> >> >> >> Memory will grow geometrically.Finally when the Old generation of
> >> JVM is
> >> >> >> Exhausted,the process will keep fgc forever. It's not an
> accident.
> >> >> >> I had blocked for 2 weeks, and today, I found it may because of an
> >> >> >> plugin:"tomcat-7.x-8.x-plugins", when the
> >> >> >> TomcatInvokeInteceptor.beforeMethod creates span,
> >> >> >> sw will set operationName by request.getRequestURI() ,in
> >> general,that
> >> >> >> should transfer to a variable such as "{userID}",but it not work.
> >> >> >> I need your help, and look forward to your suggestion
> >> >> >>
> >> >> >>
> >> >> >> Yours respectfully
> >> >> >> 2020.05.20
> >> >> >
> >> >> >--
> >> >> >Sheng Wu 吴晟
> >> >> >
> >> >> >Apache SkyWalking
> >> >> >Apache Incubator
> >> >> >Apache ShardingSphere, ECharts, DolphinScheduler podlings
> >> >> >Zipkin
> >> >> >Twitter, wusheng1108
> >> >>
> >> >>
> >> >>
> >> >>
> >> >>
> >> >>
> >>
>
Re:Re: Re: sw-bug in tomcat-7.x-8.x-plugins
Posted by 13240156934 <13...@163.com>.
Oh god,I got the answer what I wanted.
I have an suggest,that the automatically override rule should consider this situation,that if there is one or more than one aop spans between tomcat-plugins and springmvc span,it should automatically override also.
Finally ,I have two question, the one is when does the 8.0 version planned to go online.The other one is how can I know the detail that how the automatically override works.
Thanks a lot for your help,Thank you ,Thank you.
At 2020-05-21 11:17:57, "Sheng Wu" <wu...@gmail.com> wrote:
>Yes, that would break the automatically override.
>I think the better way to erase this issue, is removing the tomcat plugin,
>if you really need `com.test.rest.aop.ApiTestNetAspect.around` span,
>because I don't know what is in that span.
>
>Sheng Wu 吴晟
>Twitter, wusheng1108
>
>
>13240156934 <13...@163.com> 于2020年5月21日周四 上午11:05写道：
>
>>
>>
>>
>> All rignt,the picture may like this:
>> /v3/market/200626/candles Tomcat
>> com.test.rest.aop.ApiTestNetAspect.around
>> /v3/market/{trument_id}/candles SpringMVC
>>
>> This up 3 lines are the simple trace, you will see the there is a aop
>> between two entryspan,and the first span's url is not dynamic parameter.
>>
>>
>>
>>
>>
>>
>>
>>
>>
>>
>>
>> At 2020-05-21 10:55:33, "Sheng Wu" <wu...@gmail.com> wrote:
>> >You can't send the attachment on the mail list, provide a link to your
>> >image, please.
>> >
>> >> The second,the gc problem shouldn't happen in my prod env,because my url
>> >by dynamic parameter maybe only 10000,and costs about 10g mem is accepted.
>> >> So I wonder,if I still use sw-v7.0 and open the cache and register
>> >mechanism, and I solve the dynamic parameter problem, Is it my problem
>> that
>> >will not appear again?
>> >
>> >Once you have limited URI, and the memory cost is acceptable, then it is
>> >fine. Btw, there is no such thing `open the cache and register mechanism`.
>> >It exists there for all releases before 8.0
>> >
>> >
>> >
>> >Sheng Wu 吴晟
>> >Twitter, wusheng1108
>> >
>> >
>> >13240156934 <13...@163.com> 于2020年5月21日周四 上午10:39写道：
>> >
>> >>
>> >> hi god wu
>> >>
>> >> I have understood what you mean.It seems that's not the bug in
>> >> tomcat-7.x-8.x-plugin,it may be that I have use aop around my
>> >> controller,such as this picture.
>> >>
>> >> The second,the gc problem shouldn't happen in my prod env,because my url
>> >> by dynamic parameter maybe only 10000,and costs about 10g mem is
>> accepted.
>> >> So I wonder,if I still use sw-v7.0 and open the cache and register
>> >> mechanism, and I solve the dynamic parameter problem, Is it my problem
>> >> that will not appear again?
>> >>
>> >>
>> >>
>> >>
>> >>
>> >>
>> >> At 2020-05-20 19:18:11, "Sheng Wu" <wu...@apache.org> wrote:
>> >> >I am not sure why you say this is Tomcat plugin bug. If you are using
>> >> >spring mvc and plugin of spring mvc, the uri with parameter should be
>> >> >detected and overrided the original tomcat url automatically.
>> >> >And the GC issue is another thing, all cache and register mechanism
>> have
>> >> >been removed. No risk about that.
>> >> >
>> >> >Sheng
>> >> >
>> >> >13240156934 <13...@163.com>于2020年5月20日 周三下午5:53写道：
>> >> >
>> >> >> hi god wu.sheng:
>> >> >> First of all, thank you for your reading.
>> >> >> I have a question, and it makes us rollback the skywalking
>> deployment in
>> >> >> our prod env.And I need you help.
>> >> >> The question is ,there is lot's of url in our spring-controller,and
>> some
>> >> >> is dynamic parameter, such as "
>> http://localhost:8080/hello/user/{userID}
>> >> >> ".
>> >> >> When the qps is growing with lot's of different userID,the
>> sw-collector’s
>> >> >> Memory will grow geometrically.Finally when the Old generation of
>> JVM is
>> >> >> Exhausted,the process will keep fgc forever. It's not an accident.
>> >> >> I had blocked for 2 weeks, and today, I found it may because of an
>> >> >> plugin:"tomcat-7.x-8.x-plugins", when the
>> >> >> TomcatInvokeInteceptor.beforeMethod creates span,
>> >> >> sw will set operationName by request.getRequestURI() ,in
>> general,that
>> >> >> should transfer to a variable such as "{userID}",but it not work.
>> >> >> I need your help, and look forward to your suggestion
>> >> >>
>> >> >>
>> >> >> Yours respectfully
>> >> >> 2020.05.20
>> >> >
>> >> >--
>> >> >Sheng Wu 吴晟
>> >> >
>> >> >Apache SkyWalking
>> >> >Apache Incubator
>> >> >Apache ShardingSphere, ECharts, DolphinScheduler podlings
>> >> >Zipkin
>> >> >Twitter, wusheng1108
>> >>
>> >>
>> >>
>> >>
>> >>
>> >>
>>
Re: Re: sw-bug in tomcat-7.x-8.x-plugins
Posted by Sheng Wu <wu...@gmail.com>.
Yes, that would break the automatically override.
I think the better way to erase this issue, is removing the tomcat plugin,
if you really need `com.test.rest.aop.ApiTestNetAspect.around` span,
because I don't know what is in that span.
Sheng Wu 吴晟
Twitter, wusheng1108
13240156934 <13...@163.com> 于2020年5月21日周四 上午11:05写道：
>
>
>
> All rignt,the picture may like this:
> /v3/market/200626/candles Tomcat
> com.test.rest.aop.ApiTestNetAspect.around
> /v3/market/{trument_id}/candles SpringMVC
>
> This up 3 lines are the simple trace, you will see the there is a aop
> between two entryspan,and the first span's url is not dynamic parameter.
>
>
>
>
>
>
>
>
>
>
>
> At 2020-05-21 10:55:33, "Sheng Wu" <wu...@gmail.com> wrote:
> >You can't send the attachment on the mail list, provide a link to your
> >image, please.
> >
> >> The second,the gc problem shouldn't happen in my prod env,because my url
> >by dynamic parameter maybe only 10000,and costs about 10g mem is accepted.
> >> So I wonder,if I still use sw-v7.0 and open the cache and register
> >mechanism, and I solve the dynamic parameter problem, Is it my problem
> that
> >will not appear again?
> >
> >Once you have limited URI, and the memory cost is acceptable, then it is
> >fine. Btw, there is no such thing `open the cache and register mechanism`.
> >It exists there for all releases before 8.0
> >
> >
> >
> >Sheng Wu 吴晟
> >Twitter, wusheng1108
> >
> >
> >13240156934 <13...@163.com> 于2020年5月21日周四 上午10:39写道：
> >
> >>
> >> hi god wu
> >>
> >> I have understood what you mean.It seems that's not the bug in
> >> tomcat-7.x-8.x-plugin,it may be that I have use aop around my
> >> controller,such as this picture.
> >>
> >> The second,the gc problem shouldn't happen in my prod env,because my url
> >> by dynamic parameter maybe only 10000,and costs about 10g mem is
> accepted.
> >> So I wonder,if I still use sw-v7.0 and open the cache and register
> >> mechanism, and I solve the dynamic parameter problem, Is it my problem
> >> that will not appear again?
> >>
> >>
> >>
> >>
> >>
> >>
> >> At 2020-05-20 19:18:11, "Sheng Wu" <wu...@apache.org> wrote:
> >> >I am not sure why you say this is Tomcat plugin bug. If you are using
> >> >spring mvc and plugin of spring mvc, the uri with parameter should be
> >> >detected and overrided the original tomcat url automatically.
> >> >And the GC issue is another thing, all cache and register mechanism
> have
> >> >been removed. No risk about that.
> >> >
> >> >Sheng
> >> >
> >> >13240156934 <13...@163.com>于2020年5月20日 周三下午5:53写道：
> >> >
> >> >> hi god wu.sheng:
> >> >> First of all, thank you for your reading.
> >> >> I have a question, and it makes us rollback the skywalking
> deployment in
> >> >> our prod env.And I need you help.
> >> >> The question is ,there is lot's of url in our spring-controller,and
> some
> >> >> is dynamic parameter, such as "
> http://localhost:8080/hello/user/{userID}
> >> >> ".
> >> >> When the qps is growing with lot's of different userID,the
> sw-collector’s
> >> >> Memory will grow geometrically.Finally when the Old generation of
> JVM is
> >> >> Exhausted,the process will keep fgc forever. It's not an accident.
> >> >> I had blocked for 2 weeks, and today, I found it may because of an
> >> >> plugin:"tomcat-7.x-8.x-plugins", when the
> >> >> TomcatInvokeInteceptor.beforeMethod creates span,
> >> >> sw will set operationName by request.getRequestURI() ,in
> general,that
> >> >> should transfer to a variable such as "{userID}",but it not work.
> >> >> I need your help, and look forward to your suggestion
> >> >>
> >> >>
> >> >> Yours respectfully
> >> >> 2020.05.20
> >> >
> >> >--
> >> >Sheng Wu 吴晟
> >> >
> >> >Apache SkyWalking
> >> >Apache Incubator
> >> >Apache ShardingSphere, ECharts, DolphinScheduler podlings
> >> >Zipkin
> >> >Twitter, wusheng1108
> >>
> >>
> >>
> >>
> >>
> >>
>
Re:Re: sw-bug in tomcat-7.x-8.x-plugins
Posted by 13240156934 <13...@163.com>.
All rignt,the picture may like this:
/v3/market/200626/candles Tomcat
com.test.rest.aop.ApiTestNetAspect.around
/v3/market/{trument_id}/candles SpringMVC
This up 3 lines are the simple trace, you will see the there is a aop between two entryspan,and the first span's url is not dynamic parameter.
At 2020-05-21 10:55:33, "Sheng Wu" <wu...@gmail.com> wrote:
>You can't send the attachment on the mail list, provide a link to your
>image, please.
>
>> The second,the gc problem shouldn't happen in my prod env,because my url
>by dynamic parameter maybe only 10000,and costs about 10g mem is accepted.
>> So I wonder,if I still use sw-v7.0 and open the cache and register
>mechanism, and I solve the dynamic parameter problem, Is it my problem that
>will not appear again?
>
>Once you have limited URI, and the memory cost is acceptable, then it is
>fine. Btw, there is no such thing `open the cache and register mechanism`.
>It exists there for all releases before 8.0
>
>
>
>Sheng Wu 吴晟
>Twitter, wusheng1108
>
>
>13240156934 <13...@163.com> 于2020年5月21日周四 上午10:39写道：
>
>>
>> hi god wu
>>
>> I have understood what you mean.It seems that's not the bug in
>> tomcat-7.x-8.x-plugin,it may be that I have use aop around my
>> controller,such as this picture.
>>
>> The second,the gc problem shouldn't happen in my prod env,because my url
>> by dynamic parameter maybe only 10000,and costs about 10g mem is accepted.
>> So I wonder,if I still use sw-v7.0 and open the cache and register
>> mechanism, and I solve the dynamic parameter problem, Is it my problem
>> that will not appear again?
>>
>>
>>
>>
>>
>>
>> At 2020-05-20 19:18:11, "Sheng Wu" <wu...@apache.org> wrote:
>> >I am not sure why you say this is Tomcat plugin bug. If you are using
>> >spring mvc and plugin of spring mvc, the uri with parameter should be
>> >detected and overrided the original tomcat url automatically.
>> >And the GC issue is another thing, all cache and register mechanism have
>> >been removed. No risk about that.
>> >
>> >Sheng
>> >
>> >13240156934 <13...@163.com>于2020年5月20日 周三下午5:53写道：
>> >
>> >> hi god wu.sheng:
>> >> First of all, thank you for your reading.
>> >> I have a question, and it makes us rollback the skywalking deployment in
>> >> our prod env.And I need you help.
>> >> The question is ,there is lot's of url in our spring-controller,and some
>> >> is dynamic parameter, such as "http://localhost:8080/hello/user/{userID}
>> >> ".
>> >> When the qps is growing with lot's of different userID,the sw-collector’s
>> >> Memory will grow geometrically.Finally when the Old generation of JVM is
>> >> Exhausted,the process will keep fgc forever. It's not an accident.
>> >> I had blocked for 2 weeks, and today, I found it may because of an
>> >> plugin:"tomcat-7.x-8.x-plugins", when the
>> >> TomcatInvokeInteceptor.beforeMethod creates span,
>> >> sw will set operationName by request.getRequestURI() ,in general,that
>> >> should transfer to a variable such as "{userID}",but it not work.
>> >> I need your help, and look forward to your suggestion
>> >>
>> >>
>> >> Yours respectfully
>> >> 2020.05.20
>> >
>> >--
>> >Sheng Wu 吴晟
>> >
>> >Apache SkyWalking
>> >Apache Incubator
>> >Apache ShardingSphere, ECharts, DolphinScheduler podlings
>> >Zipkin
>> >Twitter, wusheng1108
>>
>>
>>
>>
>>
>>
Re: sw-bug in tomcat-7.x-8.x-plugins
Posted by Sheng Wu <wu...@gmail.com>.
You can't send the attachment on the mail list, provide a link to your
image, please.
> The second,the gc problem shouldn't happen in my prod env,because my url
by dynamic parameter maybe only 10000,and costs about 10g mem is accepted.
> So I wonder,if I still use sw-v7.0 and open the cache and register
mechanism, and I solve the dynamic parameter problem, Is it my problem that
will not appear again?
Once you have limited URI, and the memory cost is acceptable, then it is
fine. Btw, there is no such thing `open the cache and register mechanism`.
It exists there for all releases before 8.0
Sheng Wu 吴晟
Twitter, wusheng1108
13240156934 <13...@163.com> 于2020年5月21日周四 上午10:39写道：
>
> hi god wu
>
> I have understood what you mean.It seems that's not the bug in
> tomcat-7.x-8.x-plugin,it may be that I have use aop around my
> controller,such as this picture.
>
> The second,the gc problem shouldn't happen in my prod env,because my url
> by dynamic parameter maybe only 10000,and costs about 10g mem is accepted.
> So I wonder,if I still use sw-v7.0 and open the cache and register
> mechanism, and I solve the dynamic parameter problem, Is it my problem
> that will not appear again?
>
>
>
>
>
>
> At 2020-05-20 19:18:11, "Sheng Wu" <wu...@apache.org> wrote:
> >I am not sure why you say this is Tomcat plugin bug. If you are using
> >spring mvc and plugin of spring mvc, the uri with parameter should be
> >detected and overrided the original tomcat url automatically.
> >And the GC issue is another thing, all cache and register mechanism have
> >been removed. No risk about that.
> >
> >Sheng
> >
> >13240156934 <13...@163.com>于2020年5月20日 周三下午5:53写道：
> >
> >> hi god wu.sheng:
> >> First of all, thank you for your reading.
> >> I have a question, and it makes us rollback the skywalking deployment in
> >> our prod env.And I need you help.
> >> The question is ,there is lot's of url in our spring-controller,and some
> >> is dynamic parameter, such as "http://localhost:8080/hello/user/{userID}
> >> ".
> >> When the qps is growing with lot's of different userID,the sw-collector’s
> >> Memory will grow geometrically.Finally when the Old generation of JVM is
> >> Exhausted,the process will keep fgc forever. It's not an accident.
> >> I had blocked for 2 weeks, and today, I found it may because of an
> >> plugin:"tomcat-7.x-8.x-plugins", when the
> >> TomcatInvokeInteceptor.beforeMethod creates span,
> >> sw will set operationName by request.getRequestURI() ,in general,that
> >> should transfer to a variable such as "{userID}",but it not work.
> >> I need your help, and look forward to your suggestion
> >>
> >>
> >> Yours respectfully
> >> 2020.05.20
> >
> >--
> >Sheng Wu 吴晟
> >
> >Apache SkyWalking
> >Apache Incubator
> >Apache ShardingSphere, ECharts, DolphinScheduler podlings
> >Zipkin
> >Twitter, wusheng1108
>
>
>
>
>
>
Re:Re: sw-bug in tomcat-7.x-8.x-plugins
Posted by 13240156934 <13...@163.com>.
Add attachment picture,case it maybe lost in content.
At 2020-05-21 10:38:47, "13240156934" <13...@163.com> wrote:
hi god wu
I have understood what you mean.It seems that's not the bug in tomcat-7.x-8.x-plugin,it may be that I have use aop around my controller,such as this picture.
The second,the gc problem shouldn't happen in my prod env,because my url by dynamic parameter maybe only 10000,and costs about 10g mem is accepted.
So I wonder,if I still use sw-v7.0 and open the cache and register mechanism, and I solve the dynamic parameter problem, Is it my problem that will not appear again?
At 2020-05-20 19:18:11, "Sheng Wu" <wu...@apache.org> wrote:
>I am not sure why you say this is Tomcat plugin bug. If you are using
>spring mvc and plugin of spring mvc, the uri with parameter should be
>detected and overrided the original tomcat url automatically.
>And the GC issue is another thing, all cache and register mechanism have
>been removed. No risk about that.
>
>Sheng
>
>13240156934 <13...@163.com>于2020年5月20日 周三下午5:53写道：
>
>> hi god wu.sheng:
>> First of all, thank you for your reading.
>> I have a question, and it makes us rollback the skywalking deployment in
>> our prod env.And I need you help.
>> The question is ,there is lot's of url in our spring-controller,and some
>> is dynamic parameter, such as "http://localhost:8080/hello/user/{userID}
>> ".
>> When the qps is growing with lot's of different userID,the sw-collector’s
>> Memory will grow geometrically.Finally when the Old generation of JVM is
>> Exhausted,the process will keep fgc forever. It's not an accident.
>> I had blocked for 2 weeks, and today, I found it may because of an
>> plugin:"tomcat-7.x-8.x-plugins", when the
>> TomcatInvokeInteceptor.beforeMethod creates span,
>> sw will set operationName by request.getRequestURI() ,in general,that
>> should transfer to a variable such as "{userID}",but it not work.
>> I need your help, and look forward to your suggestion
>>
>>
>> Yours respectfully
>> 2020.05.20
>
>--
>Sheng Wu 吴晟
>
>Apache SkyWalking
>Apache Incubator
>Apache ShardingSphere, ECharts, DolphinScheduler podlings
>Zipkin
>Twitter, wusheng1108
Re: sw-bug in tomcat-7.x-8.x-plugins
Posted by 13240156934 <13...@163.com>.
hi god wu
I have understood what you mean.It seems that's not the bug in tomcat-7.x-8.x-plugin,it may be that I have use aop around my controller,such as this picture.
The second,the gc problem shouldn't happen in my prod env,because my url by dynamic parameter maybe only 10000,and costs about 10g mem is accepted.
So I wonder,if I still use sw-v7.0 and open the cache and register mechanism, and I solve the dynamic parameter problem, Is it my problem that will not appear again?
At 2020-05-20 19:18:11, "Sheng Wu" <wu...@apache.org> wrote:
>I am not sure why you say this is Tomcat plugin bug. If you are using
>spring mvc and plugin of spring mvc, the uri with parameter should be
>detected and overrided the original tomcat url automatically.
>And the GC issue is another thing, all cache and register mechanism have
>been removed. No risk about that.
>
>Sheng
>
>13240156934 <13...@163.com>于2020年5月20日 周三下午5:53写道：
>
>> hi god wu.sheng:
>> First of all, thank you for your reading.
>> I have a question, and it makes us rollback the skywalking deployment in
>> our prod env.And I need you help.
>> The question is ,there is lot's of url in our spring-controller,and some
>> is dynamic parameter, such as "http://localhost:8080/hello/user/{userID}
>> ".
>> When the qps is growing with lot's of different userID,the sw-collector’s
>> Memory will grow geometrically.Finally when the Old generation of JVM is
>> Exhausted,the process will keep fgc forever. It's not an accident.
>> I had blocked for 2 weeks, and today, I found it may because of an
>> plugin:"tomcat-7.x-8.x-plugins", when the
>> TomcatInvokeInteceptor.beforeMethod creates span,
>> sw will set operationName by request.getRequestURI() ,in general,that
>> should transfer to a variable such as "{userID}",but it not work.
>> I need your help, and look forward to your suggestion
>>
>>
>> Yours respectfully
>> 2020.05.20
>
>--
>Sheng Wu 吴晟
>
>Apache SkyWalking
>Apache Incubator
>Apache ShardingSphere, ECharts, DolphinScheduler podlings
>Zipkin
>Twitter, wusheng1108
Re: sw-bug in tomcat-7.x-8.x-plugins
Posted by Sheng Wu <wu...@apache.org>.
I am not sure why you say this is Tomcat plugin bug. If you are using
spring mvc and plugin of spring mvc, the uri with parameter should be
detected and overrided the original tomcat url automatically.
And the GC issue is another thing, all cache and register mechanism have
been removed. No risk about that.
Sheng
13240156934 <13...@163.com>于2020年5月20日 周三下午5:53写道：
> hi god wu.sheng:
> First of all, thank you for your reading.
> I have a question, and it makes us rollback the skywalking deployment in
> our prod env.And I need you help.
> The question is ,there is lot's of url in our spring-controller,and some
> is dynamic parameter, such as "http://localhost:8080/hello/user/{userID}
> ".
> When the qps is growing with lot's of different userID,the sw-collector’s
> Memory will grow geometrically.Finally when the Old generation of JVM is
> Exhausted,the process will keep fgc forever. It's not an accident.
> I had blocked for 2 weeks, and today, I found it may because of an
> plugin:"tomcat-7.x-8.x-plugins", when the
> TomcatInvokeInteceptor.beforeMethod creates span,
> sw will set operationName by request.getRequestURI() ,in general,that
> should transfer to a variable such as "{userID}",but it not work.
> I need your help, and look forward to your suggestion
>
>
> Yours respectfully
> 2020.05.20
--
Sheng Wu 吴晟
Apache SkyWalking
Apache Incubator
Apache ShardingSphere, ECharts, DolphinScheduler podlings
Zipkin
Twitter, wusheng1108