503 Error on rebuild (4.0.7)

[HighlanderICT]

We have attempted a test upgrade from our live site (3.4.8) to our dev site using 4.0.7 but we are getting a 503 error when we try to manually run the rebuild tasks on the dashboard. Has anyone else encountered this? and if so, how did you fix it? or did you? 

All our previous test upgrades encountered issues of some sort, but none had this specific issue. Going back as far as RC6 the upgrade deleted all our images from the live site (turned out to be my mistake in updating paths before running upgrade. luckily we had a backup). For the fully released product we have had issues with the archives not being added as posts (this is still not resolved but we spent 4 days unarchiving our content as a workaround), and most recently we had an issue where ip.content added html paragraph code to the title of all 1200+ articles on our site (apparently fixed in this version but not able to verify yet). Despite these issues, when it came to the rebuild process in the dashboard it completed properly each time. took a while, but no errors (503 or otherwise) and no problems afterwards that we are aware of. 

For this upgrade everything seemed to go normally and upgraded without any hitch. The process involves 3 or 4 manual sql queries that are provided, and these were a success, but that was about the only manual stuff I had to do other than make sure my paths were adjusted before i started the process. The test site and admin back end are both operational. however when accessing the admin dashboard i get a message to say the queue task is locking frequently (something that is listed as a bug fixed in this version) and the background processes to rebuild everything are frozen at the same level (0% for most). When I setup a cron job - using the instructions provided - nothing happens, and when I try to run it manually the screen starts to do something then quickly errors out with a 503 error. I logged a ticket with IPS and also with my VPS provider. My VPS provider is willing to check things out for me and are typically very quick, but as noted in countless other posts from others, IPS will not support test installs ... even when it is preventing you from updating your live install so I am in limbo .....

would appreciate it if some of my fellow IPS customers were able to offer some peer to peer help ? 

 

This is what the VPS provider noted: 

Thanks for contacting Support! I have logged into the admin page you provided, and I indeed see the 503 error you describe. When I look at the error log upon loading this page, I am seeing the following errors:

2015-06-04 02:08:09.374 [INFO] [xx.xx.xx.xx:25995-0#APVH_xxx.xxxxxxxxxxxxxx.com] connection to [/tmp/lshttpd/APVH_xxxxxxxxxxxxxx.com_Suphp.sock.602] on request #0, confirmed, 1, associated process: 30054, running: 0, error: Connection reset by peer!

2015-06-04 02:08:09.374 [NOTICE] [xx.xx.xx.xx:25995-0#APVH_xxx.xxxxxxxxxxxxxx.com] Max retries has been reached, 503!

It looks like the application is making too many connections to the server, is being cut out. Allow me a few more moments to see how to best fix this. I'll update you shortly with more information.

they then followed up with this

I apologize for the delay in reply. This appears to be an issue from Litespeed's PHP handler. Because the 503 error is general, I have been tinkering with settings that would affect the timeout. However, after making slight modifications to Litespeed >> Configuration >> Server >> External App >> LSPHP5 I have only extended the time it takes for the 503 error to appear. I have enabled Core-Dumping of your Litespeed PHP handler to better diagnose the issue - PHP keeps dying, and Litespeed's default error messages do not indicate why this is the case. Allow me some more time to better diagnose this, and I'll update you once I have more info.

and finally with this

For further troubleshooting purposes, do you know where the scripts are that the admin page is calling? 

 

It seems I cannot get this information from IPS so if anyone can help then I would be very greatful. thanks in advance. 

 

[HighlanderICT]

problem resolved. thread can be closed.