This came up before, but the thread digressed into other problems.
Centos 7 / OLS 1.5.6 / Cyberpanel 1.9
1). Wordpress sites when scanned by crawler (e.g. google, twitter) can cause a number of these errors:
No request delivery notification has been received from LSAPI application, possible dead lock.
2). The above errors seem to keep file handles open to the sockets for the specific Vhost crawled.
3). Over time the open files issue will cause OLS to error with 503s and cause an automatic restart. (to many open files)
4). The automatic restart will leave zombie threads with open file handles – openlitespeed (lshttpd - #01)
I have gone through everything I can find…
– increased the open file limit in OS
– increased the timeout for external apps in OLS (based on OLS forum and Digital Ocean recommendations)
– disabled plugins in WP
I have narrowed it down to crawler visits by IP, but it does not involve specific pages / themes / plugins or even websites.
any help would be appreciated…
rd