Was the EhCache downgrade in 1.3 intentional?

[Dan McLaughlin <dm...@tech-consortium.com>]

We've seen a significant increase in time it takes to download/upload
artifacts from Archiva after we upgraded from 1.2.2 to 1.3.  I'm in the very
beginning stages of my investigation, but one thing I just noticed when
comparing the Archiva WAR between 1.2.2 and 1.3 is that the
plexus-cache-ehcache-1.0-alpha-2.jar in the 1.3 release says it was built
March 29, 2007 by Brett...and the version in 1.2.2 was build Jun 23 2009 by
Deng.  If you JAD the EhcacheCreator classes between the two, there are some
fairly significant differences between the them.  I'm not suggesting that
this has anything to do with our issues, but I wanted to point it out as a
can't imagine that you intentionally moved back to a version of ehcache that
was two years older than the one you included in the 1.2.2 release.

Regards,

Dan

Re: Was the EhCache downgrade in 1.3 intentional?

[Deng Ching <oc...@apache.org>]

Yep, it's a connection response timeout :)

-Deng

On Wed, May 12, 2010 at 9:02 PM, Dan McLaughlin <
dmclaughlin@tech-consortium.com> wrote:

> That timeout is a connection response timeout...not a download timeout.
>  Correct?  For example, if I set it to 5 seconds I don't want large file
> downloads to start failing.
>
> -Dan
>
> On Wed, May 12, 2010 at 7:11 AM, Deng Ching <oc...@apache.org> wrote:
>
>> You can set the request timeout interval of each remote repository in the
>> repo's configuration  (from the Repositories page). The field's name is
>> 'Timeout in seconds' IIRC :)
>>
>> -Deng
>>
>>
>> On Wed, May 12, 2010 at 7:57 PM, Dan McLaughlin <
>> dmclaughlin@tech-consortium.com> wrote:
>>
>>> I think you are correct about our slowness being caused by a proxied
>>> repository that was offline.  We've seen this in the past, and its not easy
>>> to track down.  Is there a timeout/retry setting we can tune that would help
>>> reduce the impacts that an offline repository has on download performance?
>>>
>>>
>>> -Dan
>>>
>>> On Tue, May 11, 2010 at 9:15 PM, Deng Ching <oc...@apache.org> wrote:
>>>
>>>> Thanks for reporting this Dan! Good catch :)
>>>>
>>>> The downgrade in the plexus-cache-ehcache build is not intentional. It's
>>>> possible that the plexus-cache-ehcache in 1.2.2 was a patched copy in my
>>>> local repository (from a totally different issue I was working on at that
>>>> time) that was mistakenly bundled in the 1.2.2 release. But IIRC, the
>>>> changes I did was in the EhcacheCache class so I'm not sure . Since Brett
>>>> released 1.3, the plexus-cache-ehcache jar bundled with the 1.3 release
>>>> likely came from his local repository which is probably the published copy
>>>> in central.
>>>>
>>>> It might also be worthwhile to check if the remote repositories being
>>>> proxied are online. It's possible that one of them went down after the
>>>> upgrade which could also be a reason why the downloading of artifacts are
>>>> taking a while.
>>>>
>>>> -Deng
>>>>
>>>>
>>>> On Wed, May 12, 2010 at 1:09 AM, Dan McLaughlin <
>>>> dmclaughlin@tech-consortium.com> wrote:
>>>>
>>>>> We've seen a significant increase in time it takes to download/upload
>>>>> artifacts from Archiva after we upgraded from 1.2.2 to 1.3.  I'm in the
>>>>> very
>>>>> beginning stages of my investigation, but one thing I just noticed when
>>>>> comparing the Archiva WAR between 1.2.2 and 1.3 is that the
>>>>> plexus-cache-ehcache-1.0-alpha-2.jar in the 1.3 release says it was
>>>>> built
>>>>> March 29, 2007 by Brett...and the version in 1.2.2 was build Jun 23
>>>>> 2009 by
>>>>> Deng.  If you JAD the EhcacheCreator classes between the two, there are
>>>>> some
>>>>> fairly significant differences between the them.  I'm not suggesting
>>>>> that
>>>>> this has anything to do with our issues, but I wanted to point it out
>>>>> as a
>>>>> can't imagine that you intentionally moved back to a version of ehcache
>>>>> that
>>>>> was two years older than the one you included in the 1.2.2 release.
>>>>>
>>>>> Regards,
>>>>>
>>>>> Dan
>>>>>
>>>>
>>>>
>>>
>>
>

Re: Was the EhCache downgrade in 1.3 intentional?

[Dan McLaughlin <dm...@tech-consortium.com>]

That timeout is a connection response timeout...not a download timeout.
 Correct?  For example, if I set it to 5 seconds I don't want large file
downloads to start failing.

-Dan

On Wed, May 12, 2010 at 7:11 AM, Deng Ching <oc...@apache.org> wrote:

> You can set the request timeout interval of each remote repository in the
> repo's configuration  (from the Repositories page). The field's name is
> 'Timeout in seconds' IIRC :)
>
> -Deng
>
>
> On Wed, May 12, 2010 at 7:57 PM, Dan McLaughlin <
> dmclaughlin@tech-consortium.com> wrote:
>
>> I think you are correct about our slowness being caused by a proxied
>> repository that was offline.  We've seen this in the past, and its not easy
>> to track down.  Is there a timeout/retry setting we can tune that would help
>> reduce the impacts that an offline repository has on download performance?
>>
>>
>> -Dan
>>
>> On Tue, May 11, 2010 at 9:15 PM, Deng Ching <oc...@apache.org> wrote:
>>
>>> Thanks for reporting this Dan! Good catch :)
>>>
>>> The downgrade in the plexus-cache-ehcache build is not intentional. It's
>>> possible that the plexus-cache-ehcache in 1.2.2 was a patched copy in my
>>> local repository (from a totally different issue I was working on at that
>>> time) that was mistakenly bundled in the 1.2.2 release. But IIRC, the
>>> changes I did was in the EhcacheCache class so I'm not sure . Since Brett
>>> released 1.3, the plexus-cache-ehcache jar bundled with the 1.3 release
>>> likely came from his local repository which is probably the published copy
>>> in central.
>>>
>>> It might also be worthwhile to check if the remote repositories being
>>> proxied are online. It's possible that one of them went down after the
>>> upgrade which could also be a reason why the downloading of artifacts are
>>> taking a while.
>>>
>>> -Deng
>>>
>>>
>>> On Wed, May 12, 2010 at 1:09 AM, Dan McLaughlin <
>>> dmclaughlin@tech-consortium.com> wrote:
>>>
>>>> We've seen a significant increase in time it takes to download/upload
>>>> artifacts from Archiva after we upgraded from 1.2.2 to 1.3.  I'm in the
>>>> very
>>>> beginning stages of my investigation, but one thing I just noticed when
>>>> comparing the Archiva WAR between 1.2.2 and 1.3 is that the
>>>> plexus-cache-ehcache-1.0-alpha-2.jar in the 1.3 release says it was
>>>> built
>>>> March 29, 2007 by Brett...and the version in 1.2.2 was build Jun 23 2009
>>>> by
>>>> Deng.  If you JAD the EhcacheCreator classes between the two, there are
>>>> some
>>>> fairly significant differences between the them.  I'm not suggesting
>>>> that
>>>> this has anything to do with our issues, but I wanted to point it out as
>>>> a
>>>> can't imagine that you intentionally moved back to a version of ehcache
>>>> that
>>>> was two years older than the one you included in the 1.2.2 release.
>>>>
>>>> Regards,
>>>>
>>>> Dan
>>>>
>>>
>>>
>>
>

Re: Was the EhCache downgrade in 1.3 intentional?

[Deng Ching <oc...@apache.org>]

You can set the request timeout interval of each remote repository in the
repo's configuration  (from the Repositories page). The field's name is
'Timeout in seconds' IIRC :)

-Deng

On Wed, May 12, 2010 at 7:57 PM, Dan McLaughlin <
dmclaughlin@tech-consortium.com> wrote:

> I think you are correct about our slowness being caused by a proxied
> repository that was offline.  We've seen this in the past, and its not easy
> to track down.  Is there a timeout/retry setting we can tune that would help
> reduce the impacts that an offline repository has on download performance?
>
>
> -Dan
>
> On Tue, May 11, 2010 at 9:15 PM, Deng Ching <oc...@apache.org> wrote:
>
>> Thanks for reporting this Dan! Good catch :)
>>
>> The downgrade in the plexus-cache-ehcache build is not intentional. It's
>> possible that the plexus-cache-ehcache in 1.2.2 was a patched copy in my
>> local repository (from a totally different issue I was working on at that
>> time) that was mistakenly bundled in the 1.2.2 release. But IIRC, the
>> changes I did was in the EhcacheCache class so I'm not sure . Since Brett
>> released 1.3, the plexus-cache-ehcache jar bundled with the 1.3 release
>> likely came from his local repository which is probably the published copy
>> in central.
>>
>> It might also be worthwhile to check if the remote repositories being
>> proxied are online. It's possible that one of them went down after the
>> upgrade which could also be a reason why the downloading of artifacts are
>> taking a while.
>>
>> -Deng
>>
>>
>> On Wed, May 12, 2010 at 1:09 AM, Dan McLaughlin <
>> dmclaughlin@tech-consortium.com> wrote:
>>
>>> We've seen a significant increase in time it takes to download/upload
>>> artifacts from Archiva after we upgraded from 1.2.2 to 1.3.  I'm in the
>>> very
>>> beginning stages of my investigation, but one thing I just noticed when
>>> comparing the Archiva WAR between 1.2.2 and 1.3 is that the
>>> plexus-cache-ehcache-1.0-alpha-2.jar in the 1.3 release says it was built
>>> March 29, 2007 by Brett...and the version in 1.2.2 was build Jun 23 2009
>>> by
>>> Deng.  If you JAD the EhcacheCreator classes between the two, there are
>>> some
>>> fairly significant differences between the them.  I'm not suggesting that
>>> this has anything to do with our issues, but I wanted to point it out as
>>> a
>>> can't imagine that you intentionally moved back to a version of ehcache
>>> that
>>> was two years older than the one you included in the 1.2.2 release.
>>>
>>> Regards,
>>>
>>> Dan
>>>
>>
>>
>

Re: Was the EhCache downgrade in 1.3 intentional?

[Brett Porter <br...@apache.org>]

You can also take the proxy connector offline temporarily one by one.

On 12/05/2010, at 9:57 PM, Dan McLaughlin wrote:

> I think you are correct about our slowness being caused by a proxied
> repository that was offline.  We've seen this in the past, and its not easy
> to track down.  Is there a timeout/retry setting we can tune that would help
> reduce the impacts that an offline repository has on download performance?
> 
> 
> -Dan
> 
> On Tue, May 11, 2010 at 9:15 PM, Deng Ching <oc...@apache.org> wrote:
> 
>> Thanks for reporting this Dan! Good catch :)
>> 
>> The downgrade in the plexus-cache-ehcache build is not intentional. It's
>> possible that the plexus-cache-ehcache in 1.2.2 was a patched copy in my
>> local repository (from a totally different issue I was working on at that
>> time) that was mistakenly bundled in the 1.2.2 release. But IIRC, the
>> changes I did was in the EhcacheCache class so I'm not sure . Since Brett
>> released 1.3, the plexus-cache-ehcache jar bundled with the 1.3 release
>> likely came from his local repository which is probably the published copy
>> in central.
>> 
>> It might also be worthwhile to check if the remote repositories being
>> proxied are online. It's possible that one of them went down after the
>> upgrade which could also be a reason why the downloading of artifacts are
>> taking a while.
>> 
>> -Deng
>> 
>> 
>> On Wed, May 12, 2010 at 1:09 AM, Dan McLaughlin <
>> dmclaughlin@tech-consortium.com> wrote:
>> 
>>> We've seen a significant increase in time it takes to download/upload
>>> artifacts from Archiva after we upgraded from 1.2.2 to 1.3.  I'm in the
>>> very
>>> beginning stages of my investigation, but one thing I just noticed when
>>> comparing the Archiva WAR between 1.2.2 and 1.3 is that the
>>> plexus-cache-ehcache-1.0-alpha-2.jar in the 1.3 release says it was built
>>> March 29, 2007 by Brett...and the version in 1.2.2 was build Jun 23 2009
>>> by
>>> Deng.  If you JAD the EhcacheCreator classes between the two, there are
>>> some
>>> fairly significant differences between the them.  I'm not suggesting that
>>> this has anything to do with our issues, but I wanted to point it out as a
>>> can't imagine that you intentionally moved back to a version of ehcache
>>> that
>>> was two years older than the one you included in the 1.2.2 release.
>>> 
>>> Regards,
>>> 
>>> Dan
>>> 
>> 
>> 

--
Brett Porter
brett@apache.org
http://brettporter.wordpress.com/

Re: Was the EhCache downgrade in 1.3 intentional?

[Dan McLaughlin <dm...@tech-consortium.com>]

I think you are correct about our slowness being caused by a proxied
repository that was offline.  We've seen this in the past, and its not easy
to track down.  Is there a timeout/retry setting we can tune that would help
reduce the impacts that an offline repository has on download performance?


-Dan

On Tue, May 11, 2010 at 9:15 PM, Deng Ching <oc...@apache.org> wrote:

> Thanks for reporting this Dan! Good catch :)
>
> The downgrade in the plexus-cache-ehcache build is not intentional. It's
> possible that the plexus-cache-ehcache in 1.2.2 was a patched copy in my
> local repository (from a totally different issue I was working on at that
> time) that was mistakenly bundled in the 1.2.2 release. But IIRC, the
> changes I did was in the EhcacheCache class so I'm not sure . Since Brett
> released 1.3, the plexus-cache-ehcache jar bundled with the 1.3 release
> likely came from his local repository which is probably the published copy
> in central.
>
> It might also be worthwhile to check if the remote repositories being
> proxied are online. It's possible that one of them went down after the
> upgrade which could also be a reason why the downloading of artifacts are
> taking a while.
>
> -Deng
>
>
> On Wed, May 12, 2010 at 1:09 AM, Dan McLaughlin <
> dmclaughlin@tech-consortium.com> wrote:
>
>> We've seen a significant increase in time it takes to download/upload
>> artifacts from Archiva after we upgraded from 1.2.2 to 1.3.  I'm in the
>> very
>> beginning stages of my investigation, but one thing I just noticed when
>> comparing the Archiva WAR between 1.2.2 and 1.3 is that the
>> plexus-cache-ehcache-1.0-alpha-2.jar in the 1.3 release says it was built
>> March 29, 2007 by Brett...and the version in 1.2.2 was build Jun 23 2009
>> by
>> Deng.  If you JAD the EhcacheCreator classes between the two, there are
>> some
>> fairly significant differences between the them.  I'm not suggesting that
>> this has anything to do with our issues, but I wanted to point it out as a
>> can't imagine that you intentionally moved back to a version of ehcache
>> that
>> was two years older than the one you included in the 1.2.2 release.
>>
>> Regards,
>>
>> Dan
>>
>
>

Re: Was the EhCache downgrade in 1.3 intentional?

[Deng Ching <oc...@apache.org>]

Thanks for reporting this Dan! Good catch :)

The downgrade in the plexus-cache-ehcache build is not intentional. It's
possible that the plexus-cache-ehcache in 1.2.2 was a patched copy in my
local repository (from a totally different issue I was working on at that
time) that was mistakenly bundled in the 1.2.2 release. But IIRC, the
changes I did was in the EhcacheCache class so I'm not sure . Since Brett
released 1.3, the plexus-cache-ehcache jar bundled with the 1.3 release
likely came from his local repository which is probably the published copy
in central.

It might also be worthwhile to check if the remote repositories being
proxied are online. It's possible that one of them went down after the
upgrade which could also be a reason why the downloading of artifacts are
taking a while.

-Deng

On Wed, May 12, 2010 at 1:09 AM, Dan McLaughlin <
dmclaughlin@tech-consortium.com> wrote:

> We've seen a significant increase in time it takes to download/upload
> artifacts from Archiva after we upgraded from 1.2.2 to 1.3.  I'm in the
> very
> beginning stages of my investigation, but one thing I just noticed when
> comparing the Archiva WAR between 1.2.2 and 1.3 is that the
> plexus-cache-ehcache-1.0-alpha-2.jar in the 1.3 release says it was built
> March 29, 2007 by Brett...and the version in 1.2.2 was build Jun 23 2009 by
> Deng.  If you JAD the EhcacheCreator classes between the two, there are
> some
> fairly significant differences between the them.  I'm not suggesting that
> this has anything to do with our issues, but I wanted to point it out as a
> can't imagine that you intentionally moved back to a version of ehcache
> that
> was two years older than the one you included in the 1.2.2 release.
>
> Regards,
>
> Dan
>