Nightly downtime

[Lincoln]

I've been working on a vanilla installation of WordPress as part of the effort to revamp my site and I was still noticing downtimes every midnight, including last night, when the server was taking forever to respond while I was testing out my theme. I also noticed several downtimes between 3AM-5AM as well, and given the time period I'm confident that my site has been affected by these backup processes.

 

It's getting to the point now that if this doesn't resolve it once and for all that I will ask to be put on a server that does not perform backups at all, just so I can finally avoid the performance hit and downtimes, or move to another host altogether if that's not possible.

[Michael D.]

I've been working on a vanilla installation of WordPress as part of the effort to revamp my site and I was still noticing downtimes every midnight, including last night, when the server was taking forever to respond while I was testing out my theme. I also noticed several downtimes between 3AM-5AM as well, and given the time period I'm confident that my site has been affected by these backup processes.

 

It's getting to the point now that if this doesn't resolve it once and for all that I will ask to be put on a server that does not perform backups at all, just so I can finally avoid the performance hit and downtimes, or move to another host altogether if that's not possible.

I've volunteered to move you to a different server a couple of times but you've declined those offers. I'm going to split this off into it's own thread.

 

If you want to try one of the suggestions I've made, update your ticket.

[Lincoln]

I've volunteered to move you to a different server a couple of times but you've declined those offers. I'm going to split this off into it's own thread.

 

If you want to try one of the suggestions I've made, update your ticket.

 

But all those servers still perform the realsoft backups correct? Seems pointless if they are all running the same type of backup processes that's been causing these issues in the first place.

[Michael D.]

But all those servers still perform the realsoft backups correct? Seems pointless if they are all running the same type of backup processes that's been causing these issues in the first place.

I'd rather you not assume things won't be any better and just give it a try

[Michael D.]

I identified the cause for downtime at midnight - there is a user who was processing their own gzip level 9 backups at midnight (several of them, simultaneously). They run only for about a minute however since there were multiple running simultaneously they were hammering the disks hard and causing I/O requests to back up.

 

We've spaced them out so that they don't run simultaneously and will watch them tomorrow night.

[stocktrader]

I would think that customers running automatic backups would have a tendency to run them at round numbers of time, most likely at :00 and :30, with the most popular time being 12:00. It might be worth looking at the cyclical patterns of server resources. I purposely set my 2 daily backups (mySQL backup, and FTP sync backups) to run at 20 & 45 minutes past the hour to make sure it didn't add onto any load. They usually run for 3 minutes or so.

[Lincoln]

I knew it had to be a cron job/backup job running somewhere since it always happened at midnight on the dot.

 

I'm confused because I thought that CloudLinux though was supposed to insulate us from users who hog server resources, so even if they ran a bazoolion backup processes it wouldn't affect others.

 

I'll wait to see how Cypress performs once the dust settles.

[Michael D.]

It protects from CPU over-usage by PHP, CGI, etc... It doesn't insulate against disk usage such as gzip/gtar processes.