Re: Strange continous SEVERE error after upgrading to APR 1.3, Critical poller failure, Timer expired - Bayesian Filter detected spam
Hi, I just noticed that version 1.3.3 is out, but I can't see this fix in the release notes. Is the fix for this bug included in 1.3.3? Regards /Jimi mogul | jimi hullegård | system developer | hudiksvallsgatan 4, 113 30 stockholm sweden | +46 8 506 66 172 | +46 765 27 19 55 | [EMAIL PROTECTED] | www.mogul.com -Original Message- From: Nick Kew [mailto:[EMAIL PROTECTED] Sent: den 26 juni 2008 12:50 To: dev@apr.apache.org Subject: Re: Strange continous SEVERE error after upgrading to APR 1.3, Critical poller failure, Timer expired - Bayesian Filter detected spam On Thu, 26 Jun 2008 10:10:48 +0200 Jimi Hullegård [EMAIL PROTECTED] wrote: By the way, does it look like this patch (or some variant, as mentioned in other postings in this thread) will be included in a future release (like 1.3.3)? Yes. Your confirmation of the fix was what we were waiting for. Now we just have to decide which of the two variants to go with! -- Nick Kew
Re: Strange continous SEVERE error after upgrading to APR 1.3, Critical poller failure, Timer expired
On Jun 26, 2008, at 7:15 AM, Nick Kew wrote: On Wed, 25 Jun 2008 21:43:47 +0200 Ruediger Pluem [EMAIL PROTECTED] wrote: Looks good and makes sense. apr_get_netos_error() in case of ETIME = 720062 APR_TIMEUP = 70007 So they are different. which is, in itself, a little disturbing. But I would do rv = apr_get_netos_error(); instead of rv = APR_EGENERAL; Works for me in terms of fixing this bug. But looking at the history of this bug (we had a much smaller bug, had a fix, then introduced this bug by switching to apr_get_netos_error because it felt better), I'd be wary of repeating the same thought process without more thorough analysis and testing. How about APR_EGENERAL for 1.3/release, and apr_get_netos_error for trunk, where it can hopefully get some exposure? +1
RE: Strange continous SEVERE error after upgrading to APR 1.3, Critical poller failure, Timer expired
Nick Kew wrote: OK, I've just tested APR variants with Event MPM. It is indeed that fix that breaks it. When I revert the fix, the problem goes away. The original patch posted to fix the same bug works fine. Can I suggest you apply the following patch to 1.3.x (should work on 1.3.0 or 1.3.2 too) and let us know if it fixes the problem you're seeing? Index: poll/unix/port.c === --- poll/unix/port.c(revision 662044) +++ poll/unix/port.c(working copy) @@ -315,7 +315,15 @@ if (ret == -1) { (*num) = 0; -rv = apr_get_netos_error(); +if (errno == EINTR) { +rv = APR_EINTR; +} +else if (errno == ETIME) { +rv = APR_TIMEUP; +} +else { +rv = APR_EGENERAL; +} } else if (nget == 0) { rv = APR_TIMEUP; We have now applied this patch to 1.3.2 and it successfully solved the problem, ie no more errors about Critical poller failure in the log file. Thanks! By the way, does it look like this patch (or some variant, as mentioned in other postings in this thread) will be included in a future release (like 1.3.3)? Regards /Jimi
Re: Strange continous SEVERE error after upgrading to APR 1.3, Critical poller failure, Timer expired
On Thu, 26 Jun 2008 10:10:48 +0200 Jimi Hullegård [EMAIL PROTECTED] wrote: By the way, does it look like this patch (or some variant, as mentioned in other postings in this thread) will be included in a future release (like 1.3.3)? Yes. Your confirmation of the fix was what we were waiting for. Now we just have to decide which of the two variants to go with! -- Nick Kew
Re: Strange continous SEVERE error after upgrading to APR 1.3, Critical poller failure, Timer expired
On Wed, 25 Jun 2008 21:43:47 +0200 Ruediger Pluem [EMAIL PROTECTED] wrote: Looks good and makes sense. apr_get_netos_error() in case of ETIME = 720062 APR_TIMEUP = 70007 So they are different. which is, in itself, a little disturbing. But I would do rv = apr_get_netos_error(); instead of rv = APR_EGENERAL; Works for me in terms of fixing this bug. But looking at the history of this bug (we had a much smaller bug, had a fix, then introduced this bug by switching to apr_get_netos_error because it felt better), I'd be wary of repeating the same thought process without more thorough analysis and testing. How about APR_EGENERAL for 1.3/release, and apr_get_netos_error for trunk, where it can hopefully get some exposure? -- Nick Kew Application Development with Apache - the Apache Modules Book http://www.apachetutor.org/
Re: Strange continous SEVERE error after upgrading to APR 1.3, Critical poller failure, Timer expired
On 06/26/2008 01:15 PM, Nick Kew wrote: Works for me in terms of fixing this bug. But looking at the history of this bug (we had a much smaller bug, had a fix, then introduced this bug by switching to apr_get_netos_error because it felt better), I'd be wary of repeating the same thought process without more thorough analysis and testing. How about APR_EGENERAL for 1.3/release, and apr_get_netos_error for trunk, where it can hopefully get some exposure? +1. Do you commit or should I? Regards Rüdiger
Re: Strange continous SEVERE error after upgrading to APR 1.3, Critical poller failure, Timer expired
On Wed, 25 Jun 2008 10:32:58 +0200 Jimi Hullegård [EMAIL PROTECTED] wrote: Hi, After we upgraded APR from 1.2.8 to 1.3.0 we get a strange error in tomcat's catalina.out that we have never seen before: Jun 24, 2008 11:42:09 AM org.apache.tomcat.util.net.AprEndpoint$Poller run SEVERE: Critical poller failure (restarting poller): [62] Timer expired Could this be related to http://marc.info/?l=apr-devm=121347872322492w=2 I had something similar with httpd+Event MPM on solaris. Current workaround is to use Worker MPM. -- Nick Kew Application Development with Apache - the Apache Modules Book http://www.apachetutor.org/
RE: Strange continous SEVERE error after upgrading to APR 1.3, Critical poller failure, Timer expired
-Original Message- From: Nick Kew [mailto:[EMAIL PROTECTED] Sent: den 25 juni 2008 13:44 To: dev@apr.apache.org Subject: Re: Strange continous SEVERE error after upgrading to APR 1.3, Critical poller failure, Timer expired On Wed, 25 Jun 2008 10:32:58 +0200 Jimi Hullegård [EMAIL PROTECTED] wrote: Hi, After we upgraded APR from 1.2.8 to 1.3.0 we get a strange error in tomcat's catalina.out that we have never seen before: Jun 24, 2008 11:42:09 AM org.apache.tomcat.util.net.AprEndpoint$Poller run SEVERE: Critical poller failure (restarting poller): [62] Timer expired Could this be related to http://marc.info/?l=apr-devm=121347872322492w=2 I had something similar with httpd+Event MPM on solaris. Current workaround is to use Worker MPM. I'm not quite sure I follow you now... Worker MPM is for Apache Httpd, right? We don't use httpd. Do you know of any workaround for Tomcat? /Jimi
Re: Strange continous SEVERE error after upgrading to APR 1.3, Critical poller failure, Timer expired
On Wed, 2008-06-25 at 12:43, Nick Kew wrote: On Wed, 25 Jun 2008 10:32:58 +0200 Jimi Hullegård [EMAIL PROTECTED] wrote: Hi, After we upgraded APR from 1.2.8 to 1.3.0 we get a strange error in tomcat's catalina.out that we have never seen before: Jun 24, 2008 11:42:09 AM org.apache.tomcat.util.net.AprEndpoint$Poller run SEVERE: Critical poller failure (restarting poller): [62] Timer expired Could this be related to http://marc.info/?l=apr-devm=121347872322492w=2 I had something similar with httpd+Event MPM on solaris. Current workaround is to use Worker MPM. OK, I've just tested APR variants with Event MPM. It is indeed that fix that breaks it. When I revert the fix, the problem goes away. The original patch posted to fix the same bug works fine. Can I suggest you apply the following patch to 1.3.x (should work on 1.3.0 or 1.3.2 too) and let us know if it fixes the problem you're seeing? Index: poll/unix/port.c === --- poll/unix/port.c(revision 662044) +++ poll/unix/port.c(working copy) @@ -315,7 +315,15 @@ if (ret == -1) { (*num) = 0; -rv = apr_get_netos_error(); +if (errno == EINTR) { +rv = APR_EINTR; +} +else if (errno == ETIME) { +rv = APR_TIMEUP; +} +else { +rv = APR_EGENERAL; +} } else if (nget == 0) { rv = APR_TIMEUP; -- Nick Kew
RE: Strange continous SEVERE error after upgrading to APR 1.3, Critical poller failure, Timer expired
Nick Kew wrote: On Wed, 2008-06-25 at 12:43, Nick Kew wrote: Could this be related to http://marc.info/?l=apr-devm=121347872322492w=2 I had something similar with httpd+Event MPM on solaris. Current workaround is to use Worker MPM. OK, I've just tested APR variants with Event MPM. It is indeed that fix that breaks it. When I revert the fix, the problem goes away. The original patch posted to fix the same bug works fine. Can I suggest you apply the following patch to 1.3.x (should work on 1.3.0 or 1.3.2 too) and let us know if it fixes the problem you're seeing? [patch code snipped] Ok, I have forwarded this to some persons that have access to the system now (I am currently at a different office with no remote access). I hope they have time to test this soon. /Jimi
Re: Strange continous SEVERE error after upgrading to APR 1.3, Critical poller failure, Timer expired
On 06/25/2008 02:55 PM, Nick Kew wrote: On Wed, 2008-06-25 at 12:43, Nick Kew wrote: On Wed, 25 Jun 2008 10:32:58 +0200 Jimi Hullegård [EMAIL PROTECTED] wrote: Hi, After we upgraded APR from 1.2.8 to 1.3.0 we get a strange error in tomcat's catalina.out that we have never seen before: Jun 24, 2008 11:42:09 AM org.apache.tomcat.util.net.AprEndpoint$Poller run SEVERE: Critical poller failure (restarting poller): [62] Timer expired Could this be related to http://marc.info/?l=apr-devm=121347872322492w=2 I had something similar with httpd+Event MPM on solaris. Current workaround is to use Worker MPM. OK, I've just tested APR variants with Event MPM. It is indeed that fix that breaks it. When I revert the fix, the problem goes away. The original patch posted to fix the same bug works fine. Can I suggest you apply the following patch to 1.3.x (should work on 1.3.0 or 1.3.2 too) and let us know if it fixes the problem you're seeing? Index: poll/unix/port.c === --- poll/unix/port.c(revision 662044) +++ poll/unix/port.c(working copy) @@ -315,7 +315,15 @@ if (ret == -1) { (*num) = 0; -rv = apr_get_netos_error(); +if (errno == EINTR) { +rv = APR_EINTR; +} +else if (errno == ETIME) { +rv = APR_TIMEUP; +} +else { +rv = APR_EGENERAL; +} } else if (nget == 0) { rv = APR_TIMEUP; Looks good and makes sense. apr_get_netos_error() in case of ETIME = 720062 APR_TIMEUP = 70007 So they are different. But I would do rv = apr_get_netos_error(); instead of rv = APR_EGENERAL; Regards Rüdiger
Re: Strange continous SEVERE error after upgrading to APR 1.3, Critical poller failure, Timer expired
Ruediger Pluem wrote: On 06/25/2008 02:55 PM, Nick Kew wrote: OK, I've just tested APR variants with Event MPM. It is indeed that fix that breaks it. When I revert the fix, the problem goes away. The original patch posted to fix the same bug works fine. Can I suggest you apply the following patch to 1.3.x (should work on 1.3.0 or 1.3.2 too) and let us know if it fixes the problem you're seeing? Index: poll/unix/port.c === --- poll/unix/port.c(revision 662044) +++ poll/unix/port.c(working copy) @@ -315,7 +315,15 @@ if (ret == -1) { (*num) = 0; -rv = apr_get_netos_error(); +if (errno == EINTR) { +rv = APR_EINTR; +} +else if (errno == ETIME) { +rv = APR_TIMEUP; +} +else { +rv = APR_EGENERAL; +} } else if (nget == 0) { rv = APR_TIMEUP; Looks good and makes sense. apr_get_netos_error() in case of ETIME = 720062 APR_TIMEUP = 70007 So they are different. But I would do rv = apr_get_netos_error(); instead of rv = APR_EGENERAL; Good observations. Folks, you want to fix these regressions before we TR 1.3.3? Bill