You are viewing a plain text version of this content. The canonical link for it is here.
Posted to users@tomcat.apache.org by Stephane Passignat <pa...@hotmail.com> on 2022/07/06 11:12:32 UTC
Re: Tomcat freezes with axios
Hi,
Thanks for your help. I found that someone else was freezing the API
server through the database (very long running uncommitted transactions).
Anyway I also limited the number of parallel connections on javascript
side (axios).
regards,
Stephane
Le 2022-06-30 à 18:42, Christopher Schultz a écrit :
> All,
>
> On 6/30/22 02:34, Mark Thomas wrote:
>> Hi,
>>
>> We need more information to help you.
>>
>> Tomcat version?
>>
>> Tomcat connector configuration (from server.xml)?
>>
>> httpd version?
>>
>> httpd MPM and configuration?
>>
>> mod_proxy configuration?
>>
>> Was the httpd restart graceful or not?
>
> Wild guess: missing finally { java.sql.Connection.close(); }
>
> -chris
>
>> On 29/06/2022 19:36, Stephane Passignat wrote:
>>> Hello,
>>>
>>> I'm creating a SAP application performing REST call on an API
>>> running on Tomcat. Tomcat runs behind an apache reverse-proxy and
>>> communication between them use http. The calls are executed with
>>> axios using a basic authentication.
>>>
>>>
>>> Everything runs fine for a moment, but for an unknown reason all
>>> http request are hanging after some time and hundreds or maybe
>>> thousands requests (if these metrics make any sense).
>>>
>>>
>>> In chrome, the requests are in a 'pending' status.
>>>
>>> Restarting chrome allows to do one or two requests and then issue
>>> occurs again
>>>
>>> Restarting apache doesn't change anything.
>>>
>>> Restarting Tomcat resolve the situation. Tomcat shutdow is a bit
>>> longer. Request in chrome ends when tomcat stops.
>>>
>>>
>>> I'm not very inspired by this issue and actually trying to find
>>> inspiration in jmx and log files but nothing pops up.
>>>
>>>
>>> Does someone have an idea ?
>>>
>>>
>>>
>>> thanks
>>>
>>> Stephane
>>>
>>>
>>> ---------------------------------------------------------------------
>>> To unsubscribe, e-mail: users-unsubscribe@tomcat.apache.org
>>> For additional commands, e-mail: users-help@tomcat.apache.org
>>>
>>
>> ---------------------------------------------------------------------
>> To unsubscribe, e-mail: users-unsubscribe@tomcat.apache.org
>> For additional commands, e-mail: users-help@tomcat.apache.org
>>
>
> ---------------------------------------------------------------------
> To unsubscribe, e-mail: users-unsubscribe@tomcat.apache.org
> For additional commands, e-mail: users-help@tomcat.apache.org
>
---------------------------------------------------------------------
To unsubscribe, e-mail: users-unsubscribe@tomcat.apache.org
For additional commands, e-mail: users-help@tomcat.apache.org
Re: Tomcat freezes with axios
Posted by Stephane Passignat <pa...@hotmail.com>.
Hi Christopher,
I'm agreeing with you about long transaction for a general JEE
application design. But that's probably the one case over a million.
Tomcat is used as the middle tiers server for a jdbc type 3 driver, with
an IDEs as a front-end. In this situation we know transaction can be
longer than application ones (if we exclude conversation patterns).
This solution helps us a lot to enforce security on "direct" databases
accesses (maintenance, data fixing, ...). It also provides a very
usefull audit trail.
Transactions were really too long because some users forgot to commit
(or rollback) their work, on several parallel connections... As I was
doing the first runs of a new JS application at the same time, I though
it was this new application.
Stephane
Le 2022-07-06 à 23:37, Christopher Schultz a écrit :
> Stephane,
>
> On 7/6/22 07:12, Stephane Passignat wrote:
>> Thanks for your help. I found that someone else was freezing the API
>> server through the database (very long running uncommitted
>> transactions).
>
> That'll definitely do it.
>
> This kind of thing isn't going to work in the real-world: tying-up a
> database with a transaction that lasts even a few seconds is going to
> absolutely kill your performance.
>
> You need to work with that other group to figure out why they need
> long-running transactions and figure out how to solve the problem in a
> different way.
>
> You should also protect yourself and set timeouts on writes that, if
> they fail, they fail "fast" and don't appear to freeze. You should be
> able to detect a write-timeout and reply to the API user saying
> "sorry, write failed after N ms" or something like that. Otherwise,
> you run the risk of a single uncommitted transaction stalling your
> entire business while writes pile-up. Users tired of waiting will
> re-request the same write over and over again. When the initial
> transaction finally commits, you'll have a huge storm of writes
> hitting your database all at the same time. It will be a mess.
>
>> Anyway I also limited the number of parallel connections on
>> javascript side (axios).
>
> This is always an excellent idea. There is no reason for a single
> client to be making huge numbers of queries to your database
> simultaneously.
>
> -chris
>
>> Le 2022-06-30 à 18:42, Christopher Schultz a écrit :
>>> All,
>>>
>>> On 6/30/22 02:34, Mark Thomas wrote:
>>>> Hi,
>>>>
>>>> We need more information to help you.
>>>>
>>>> Tomcat version?
>>>>
>>>> Tomcat connector configuration (from server.xml)?
>>>>
>>>> httpd version?
>>>>
>>>> httpd MPM and configuration?
>>>>
>>>> mod_proxy configuration?
>>>>
>>>> Was the httpd restart graceful or not?
>>>
>>> Wild guess: missing finally { java.sql.Connection.close(); }
>>>
>>> -chris
>>>
>>>> On 29/06/2022 19:36, Stephane Passignat wrote:
>>>>> Hello,
>>>>>
>>>>> I'm creating a SAP application performing REST call on an API
>>>>> running on Tomcat. Tomcat runs behind an apache reverse-proxy and
>>>>> communication between them use http. The calls are executed with
>>>>> axios using a basic authentication.
>>>>>
>>>>>
>>>>> Everything runs fine for a moment, but for an unknown reason all
>>>>> http request are hanging after some time and hundreds or maybe
>>>>> thousands requests (if these metrics make any sense).
>>>>>
>>>>>
>>>>> In chrome, the requests are in a 'pending' status.
>>>>>
>>>>> Restarting chrome allows to do one or two requests and then issue
>>>>> occurs again
>>>>>
>>>>> Restarting apache doesn't change anything.
>>>>>
>>>>> Restarting Tomcat resolve the situation. Tomcat shutdow is a bit
>>>>> longer. Request in chrome ends when tomcat stops.
>>>>>
>>>>>
>>>>> I'm not very inspired by this issue and actually trying to find
>>>>> inspiration in jmx and log files but nothing pops up.
>>>>>
>>>>>
>>>>> Does someone have an idea ?
>>>>>
>>>>>
>>>>>
>>>>> thanks
>>>>>
>>>>> Stephane
>>>>>
>>>>>
>>>>> ---------------------------------------------------------------------
>>>>> To unsubscribe, e-mail: users-unsubscribe@tomcat.apache.org
>>>>> For additional commands, e-mail: users-help@tomcat.apache.org
>>>>>
>>>>
>>>> ---------------------------------------------------------------------
>>>> To unsubscribe, e-mail: users-unsubscribe@tomcat.apache.org
>>>> For additional commands, e-mail: users-help@tomcat.apache.org
>>>>
>>>
>>> ---------------------------------------------------------------------
>>> To unsubscribe, e-mail: users-unsubscribe@tomcat.apache.org
>>> For additional commands, e-mail: users-help@tomcat.apache.org
>>>
---------------------------------------------------------------------
To unsubscribe, e-mail: users-unsubscribe@tomcat.apache.org
For additional commands, e-mail: users-help@tomcat.apache.org
Re: Tomcat freezes with axios
Posted by Christopher Schultz <ch...@christopherschultz.net>.
Stephane,
On 7/6/22 07:12, Stephane Passignat wrote:
> Thanks for your help. I found that someone else was freezing the API
> server through the database (very long running uncommitted transactions).
That'll definitely do it.
This kind of thing isn't going to work in the real-world: tying-up a
database with a transaction that lasts even a few seconds is going to
absolutely kill your performance.
You need to work with that other group to figure out why they need
long-running transactions and figure out how to solve the problem in a
different way.
You should also protect yourself and set timeouts on writes that, if
they fail, they fail "fast" and don't appear to freeze. You should be
able to detect a write-timeout and reply to the API user saying "sorry,
write failed after N ms" or something like that. Otherwise, you run the
risk of a single uncommitted transaction stalling your entire business
while writes pile-up. Users tired of waiting will re-request the same
write over and over again. When the initial transaction finally commits,
you'll have a huge storm of writes hitting your database all at the same
time. It will be a mess.
> Anyway I also limited the number of parallel connections on javascript
> side (axios).
This is always an excellent idea. There is no reason for a single client
to be making huge numbers of queries to your database simultaneously.
-chris
> Le 2022-06-30 à 18:42, Christopher Schultz a écrit :
>> All,
>>
>> On 6/30/22 02:34, Mark Thomas wrote:
>>> Hi,
>>>
>>> We need more information to help you.
>>>
>>> Tomcat version?
>>>
>>> Tomcat connector configuration (from server.xml)?
>>>
>>> httpd version?
>>>
>>> httpd MPM and configuration?
>>>
>>> mod_proxy configuration?
>>>
>>> Was the httpd restart graceful or not?
>>
>> Wild guess: missing finally { java.sql.Connection.close(); }
>>
>> -chris
>>
>>> On 29/06/2022 19:36, Stephane Passignat wrote:
>>>> Hello,
>>>>
>>>> I'm creating a SAP application performing REST call on an API
>>>> running on Tomcat. Tomcat runs behind an apache reverse-proxy and
>>>> communication between them use http. The calls are executed with
>>>> axios using a basic authentication.
>>>>
>>>>
>>>> Everything runs fine for a moment, but for an unknown reason all
>>>> http request are hanging after some time and hundreds or maybe
>>>> thousands requests (if these metrics make any sense).
>>>>
>>>>
>>>> In chrome, the requests are in a 'pending' status.
>>>>
>>>> Restarting chrome allows to do one or two requests and then issue
>>>> occurs again
>>>>
>>>> Restarting apache doesn't change anything.
>>>>
>>>> Restarting Tomcat resolve the situation. Tomcat shutdow is a bit
>>>> longer. Request in chrome ends when tomcat stops.
>>>>
>>>>
>>>> I'm not very inspired by this issue and actually trying to find
>>>> inspiration in jmx and log files but nothing pops up.
>>>>
>>>>
>>>> Does someone have an idea ?
>>>>
>>>>
>>>>
>>>> thanks
>>>>
>>>> Stephane
>>>>
>>>>
>>>> ---------------------------------------------------------------------
>>>> To unsubscribe, e-mail: users-unsubscribe@tomcat.apache.org
>>>> For additional commands, e-mail: users-help@tomcat.apache.org
>>>>
>>>
>>> ---------------------------------------------------------------------
>>> To unsubscribe, e-mail: users-unsubscribe@tomcat.apache.org
>>> For additional commands, e-mail: users-help@tomcat.apache.org
>>>
>>
>> ---------------------------------------------------------------------
>> To unsubscribe, e-mail: users-unsubscribe@tomcat.apache.org
>> For additional commands, e-mail: users-help@tomcat.apache.org
>>
---------------------------------------------------------------------
To unsubscribe, e-mail: users-unsubscribe@tomcat.apache.org
For additional commands, e-mail: users-help@tomcat.apache.org