Web Hosting Forum - Net Hosting Talk

We are a community of individuals and businesses passionate about web hosting. Let's build, learn, and grow together.

apache crash several times a day



I have a server running Debian hosted at OVH with several websites.

For several months, between 5 and 20 times a day, apache becomes inaccessible for 30 to 60 seconds and then runs again.
And so for our customers, the site oar until the end of the cut ...

I first tested the other services which are all accessible without problem: SSH, FTP, mails

Concerning apache logs, they literally stop during the break and nothing appears.

Concerning the health of the server during these cuts after Munin and NetData:
- Apache: all graphs stop recording completely during power down, then activity resumes after
- Server activity is normal and far from overloaded (CPU, RAM, Disk)
- No significant peak in attendance
- No script running

I do not think that a redirection error or in a PHP script could cause this kind of crash but I may be wrong (and this would be a priori visible in the logs).

This problem was detected shortly after a change of server and the passage of our site in HTTPS but we found nothing that suggests that it comes directly from this ...

After testing many things with our outsourcer and our OVH contact, we still have not figured out how to solve our problem.
The server's motherboard and memory have been changed to exclude a hardware problem.

If you have any track I am taker !!
Thank you in advance.

For further information:
Hardware configuration:
- CPU: Intel (R) Xeon (R) CPU E5-2650 v3 @ 2.30GHz 40 cores
- RAM: 8x32GB
- Discs: 4x800GB SSD
Software Configuration:
- Debian
- Apache 2.4.27
- Php 7.0.23
  • Advertisement
  • Our last attempts were to disable the logs one by one to make sure that one of them was not a problem but without success. Then our outsourcer deactivated user and group quota files and there: no more hangs for 24h.

    Since the re-creation and replenishment of quota files, more blockages, no more problems, probably that one of them was just corrupted ...

    I hope this topic will serve others :)
  • Advertisement
  • Advertisement