UPS-398 - September 21st Easyapache update causing segfaults in apache/PHP 5.6
We have seen issues on a number of servers this morning with apache with PHP 5.6 crashing since last night's upcp update. The affected package appears to be ea-apache24-2.4.49-1.1.1.cpanel.x86_64, so we have rolled back to the previous version and frozen updates.
-
Don't know about PHP 5.6 specifically, but I'm also having problems with apache 2.4.49. After about 1.5 hours, apache stops responding completely. Restarting apache brings it back for another 1.5 hours or so. I run PHP versions from 5.6 to 7.4 via PHP-FPM which remains functional once apache is restarted. I also downgraded all the ea-apache24 related updates but added a cron firing ever 15 minutes to graceful restart just in case since I don't want to risk their still being an issue when I'm asleep. Googling shows some Linux and DirectAdmin users experiencing similar issues with 2.4.49, but some are fine. 0 -
After about 1.5 hours, apache stops responding completely. Restarting apache brings it back for another 1.5 hours or so. I run PHP versions from 5.6 to 7.4 via PHP-FPM which remains functional once apache is restarted.
Exactly the same situation, although it seems that only 20% of our servers are affected by this issue. I hope cPanel finds a solution soon, should we contact support?0 -
That's quite interesting - do that 20% have Cloudlinux and the others don't, or the other way around? Exactly the same situation, although it seems that only 20% of our servers are affected by this issue. I hope cPanel finds a solution soon, should we contact support?
0 -
We do not use Cloudlinux. It only happens on virtualized servers (virtuozzo and kvm), but not all virtualized ones are affected, very strange. All non-virtualized servers work fine. 0 -
We're seeing it on physical servers. It is clear there is problem. I admin a bunch of cPanel servers but we aren't a registered user or reseller and I'm not sure whether we can raise a ticket without a licence so it may be easier for someone else to do it. We do not use Cloudlinux. It only happens on virtualized servers (virtuozzo and kvm), but not all virtualized ones are affected, very strange. All non-virtualized servers work fine.
0 -
What mpm are you using? ######## rpm -qa|grep mpm ######## I had a similar case and it was event MPM and switching to prefork did make things better 0 -
Hey everyone - we're definitely aware, although it is an upstream issue with Apache. We're tracking it through an internal upstream case as it is related to this: but since this is still so new we haven't tested on our end just yet. We're working on it though and I'll post an update once I have more details. 0 -
Update - we've pulled apache24-2.4.49 from our mirrors so the problematic update will not effect systems that have not updated yet. 0 -
Looks like Apache 2.4.49 modules are still being listed in the repository. Although Apache 2.4.49 itself is not listed. I'm not sure if that would be problematic or not. Also of note - updating cPanel also runs a yum update. There was a new cPanel update a few days ago (at least for the 94 branch). So even if you have automatic yum updates disabled, if you update cPanel, yum update is going to run. This has been problematic for me in the past when a cPanel update and an EasyApache update is released on the same day (EA updates are usually done on Thursdays it would seem). Although this may be mitigated since Apache 2.4.49 itself has been removed from the repository for right now. I mention this because someone may have been aware of the Apache 2.4.49 issues and wanted to avoid it, but if they run upcp, then they will inadvertently get updated to Apache 2.4.49 (at least prior to cPanel removing Apache 2.4.49 from the repository). 0 -
As a workaround in the meantime, I switched Apache from mpm_event to mpm_worker Which seems to be working until a patch. 0 -
As a workaround in the meantime, I switched Apache from mpm_event to mpm_worker Which seems to be working until a patch.
Tested it on two separate machines; still saw 1-2 mpm_worker SIGTERMs, but only right after swapping. Both have been good for about 45 minutes now, instead of having to restart httpd every 10-15 minutes.0 -
Update - we've pulled apache24-2.4.49 from our mirrors so the problematic update will not effect systems that have not updated yet.
@cPRex Is there a way to roll back the update? -Michael0 -
Currently we are recommending that you switch to mpm_worker as outlined here: cPanel We don't recommend trying to downgrade the packages are there are several that are affected - my personal system has 13 dependencies for ea-apache24 when I checked just now. We're already working on a fix on our side and hope to have that out soon. 0 -
Currently we are recommending that you switch to mpm_worker as outlined here: https://i.imgur.com/h0spTQF.png
https://i.imgur.com/oCiCidP.png
Is this by any chance because y'all removed 2.4.49 from your mirrors? Or something else? -Michael0 -
I wouldn't expect that to be caused by the mirror changes. If you're seeing odd issues with EA4, could you submit a ticket to our team so we can check that? 0 -
I wouldn't expect that to be caused by the mirror changes. If you're seeing odd issues with EA4, could you submit a ticket to our team so we can check that?
Based on average reply times and the fact that I am just doing this as a workaround to something the team is actively trying to fix, do you think it's worth submitting on? Ie. could this be indicative of some other underlying issue that should be looked at? Or should I wait until the fix comes out? -Michael0 -
I can say I haven't run into other uses having issues switching the mpm today. The queue isn't so bad right now if you wanted to put in a ticket, but it seems to me like this is a separate issue from the main Apache problems. 0 -
Hello, I have same issue on Ubuntu 20 since last apache 2.4.49 update official update. I dont use cpanel. No error trace in logs and solved after apache2 service restart. The problem recurs every 4 to 6 hours depending on the traffic. Hope that can help to identify issue and solve it. 0 -
Apache not stable after update to 2.4.49-1.1.1.cpanel.x86_64 https://assets.zendesk.com/hc/assets/default_avatar.png
Jeremy Maldonado- 6 hours ago
- Updated
0 -
@Spirogg Thank you, do you know whether or not a restart is required after updating in order to apply the patch? -Michael 0 -
@Spirogg Thank you, do you know whether or not a restart is required after updating in order to apply the patch? -Michael
I'm not sure but I just did the update now without restart? I am wondering the same thing. I guess we can wait and see if the issue persists after an hour or so ? I don't have much on the server so I can restart if need be but i wanted to see if it worked without restart for another server :) sorry I don't have that answer :( Regards Spiro0 -
I ran the update, but when I search the log I don't see any messages referencing Apache, so I am not sure how to tell if it was applied or what the difference is. The Apache version number is the same, although it does say that it was built today: [root@vps2 ~]# httpd -v Server version: Apache/2.4.49 (cPanel) Server built: Sep 23 2021 18:55:07
@cPRex, do you know how to tell whether or not the patch was applied? Thanks. Edit: Ohhh, it did fix the other issue where I couldn't change from mpm_event to mpm_worker, so something is fixed at least. :D -Michael0 -
I ran the update, but when I search the log I don't see any messages referencing Apache, so I am not sure how to tell if it was applied or what the difference is. The Apache version number is the same, although it does say that it was built today:
[root@vps2 ~]# httpd -v Server version: Apache/2.4.49 (cPanel) Server built: Sep 23 2021 18:55:07
@cPRex, do you know how to tell whether or not the patch was applied? Thanks. Edit: Ohhh, it did fix the other issue where I couldn't change from mpm_event to mpm_worker, so something is fixed at least. :D -Michael
hi not sure how to tell. I just ran the update as you did and seems it was fixed.. One thing I did notice is I did yum update as well and I did update the kernel. I do not use Kernelcare so i updated anyway. then WHM asked me to restart server so I did it. then when I clicked to return to WHM and login after 1 minute or so, It said my license was not valid. so after a couple more minutes of refresh and closed browser and then was able to login again?? not sure why that happened after the restart but just a heads up. as I said I was finally able to login after about 3 mins of server coming back online. just a little weird, but all is good now. Sorry I do not have enough to tell you how to double check to update. but I do not see any issues so far on my end and you were able to switch mpm so as you said something is fixed :) regards Spiro0 -
You would need to check the full RPM version to ensure the update was applied, but no server restart is necessary. 0
Please sign in to leave a comment.
Comments
25 comments