OK, here’s a problem that we\\\’ve been experiencing, wonder if anyone can help us on. I’ll try to keep it brief but I want to provide as much info as possible:
We run 3 Magento websites (Magento Professional Ed.) AND our MySQL server on one dedicated Peer1 server with the following specs:
-Red Hat Enterprise Linux 5
-2 x Intel Xeon E5620 2.50 GHz Quad-Core Processor
-24GB 1333 Mhz DDR3 RAM
-4 x 146GB 15 SAS drives in RAID 10
-Tivoli Managed Backups
For all three of our websites COMBINED, we have the following 2012 website visitor metrics:
---2012 Non Peak Months (February-July & September-December): 120,341 visitors (12,034/mo over a 10-month average)
---2012 Peak Periods: January: 22,006 visitors and August: 29,119 visitors.
We have 6,000 Item SKUs.
We’ve had little problem with Magento until, suddenly, in early Dec. 2012, we noticed that whoever tried to initially log into our Magento Admin Panel would experience a time-out. Second attempt to log-in would generally work fine. Error message displayed: SQLSTATE[HY000]: General error: 2006 MySQL server has gone away.
This log-in timeout happens every morning. Slowly, it came to our attention that not only was this timeout problem occurring during admin panel log-in attempt, but at other points as well, including customer website searches, had it timeout when saving an item in admin panel, and, most importantly, time-outs after a customer submits an order from checkout.
Developer told us from logs, looks like error happens anywhere from 5 to 20 times daily. Aside from regular early morning admin panel timeout, it is hard to replicate on demand, but when it happens, it seems to fix itself shortly afterwards, only to fail again later.
Researched error message on Internet: virtually all posts we came across indicated it has to do with MySQL settings. However, we optimized these settings for Magento up and down, (including wait_timeout and max_allowed_packet). Did not fix it. We implemented Cloudflare to reduce server load and regulate malicious traffic. Did not fix it. Cleared out large MySQL logs. Did not fix it. We migrated from our basic MySQL to more stable Percona MySQL. Did not fix it. Database admin at server company said, in these instances, 80% of time error is with Magento, not database. We did not believe him until we exhausted all of the above database fix attempts with no luck.
At this point, our websites are still unstable, timing out randomly on customer orders, etc. Although, the timeout error to access Magento Admin Panel still occurs regularly on the intial log-in attempt each morning.
Has anyone else experienced this or any other ideas that might help us solve this (aside from the solution attempts I indicated above that we have already tried)? At this point, anything would be appreciated.