Frequently Asked Question
Gateway timeouts?
Last Updated 5 years ago
Gateway timeouts are when the Webserver process gives up on waiting for the PHP process that it requested. With us regularly balancing server load by moving sites on busy servers to new ones, for instance timeouts at 1-8AM on a Sunday here in in the states, is not exactly a busy time in regards to high load.
If you have several sites with us, most of them may be on the same server. So if only one is experiencing weirdness, it's likely something with that site and not the server or hosting.
So if something is going on, it's going to be one of three things:
1) Something wrong with our hosting
2) Something wrong internal to your site
3) Something wrong with an external site yours depends on.
--------
1)
Gateway timeout errors are usually due to the third option. If your site needs something from a remote server, then our PHP process will make a connection to that server and wait on the results. But, if that's still going when the Webserver process hits its timeout, the Webserver process will close the connection, and show a gateway timeout error.
This is normally the issue - Akismet and various other plugins will call home, and wait for info.
2)
Next most often is a cluttered database. Have a million+ rows of info in your database? The problem is that plugin writers are terrible at using the database, and don't put a whit of energy into cleanup or searching for info efficiently.
Think of a random list of 20 names. It doesn't take that long to sort them in alphabetical order, so it doesn't really matter what method you use to sort them. But with a list of 1 million names, it can make a tremendous difference in time spent.
In general, we recommend that clients purge old clients and sales info - especially that of $0 sales. It just leaves a mess.
We do our best by trying to keep the database's working set in memory, but there is only so much we can do to work around poor plugins.
If you have someone on retainer see if you can get them to clean up the database.
3)
Hosting issues, and things which can be fixed at the hosting tier.
Server load is the big one here. On our slower days you shouldn't worry. Like on Sundays, and even then we have really fast solid state storage drives for your site(s).
One thing to note is if you are using the really-simple-ssl plugin. It used to be fantastic, but recent versions of it are getting slower and slower. Ask us to convert your site for ssl/HTTPS for $25, and we can eliminate that as a possibility.
Our "normal" setting for timeouts is 30 seconds. Should you have an issue or specific request requiring a longer timeout to be applied we do have an experimental setting to extend that time. Extending it does make us more vulnerable to various "Denial of Service" attacks where the bozos keep opening connections to the server to tie up all the resources they can. If we do apply that for a circumstance, we also reserve the right to move it back to 30 seconds should we deem it necessary.
If you have several sites with us, most of them may be on the same server. So if only one is experiencing weirdness, it's likely something with that site and not the server or hosting.
So if something is going on, it's going to be one of three things:
1) Something wrong with our hosting
2) Something wrong internal to your site
3) Something wrong with an external site yours depends on.
--------
1)
Gateway timeout errors are usually due to the third option. If your site needs something from a remote server, then our PHP process will make a connection to that server and wait on the results. But, if that's still going when the Webserver process hits its timeout, the Webserver process will close the connection, and show a gateway timeout error.
This is normally the issue - Akismet and various other plugins will call home, and wait for info.
2)
Next most often is a cluttered database. Have a million+ rows of info in your database? The problem is that plugin writers are terrible at using the database, and don't put a whit of energy into cleanup or searching for info efficiently.
Think of a random list of 20 names. It doesn't take that long to sort them in alphabetical order, so it doesn't really matter what method you use to sort them. But with a list of 1 million names, it can make a tremendous difference in time spent.
In general, we recommend that clients purge old clients and sales info - especially that of $0 sales. It just leaves a mess.
We do our best by trying to keep the database's working set in memory, but there is only so much we can do to work around poor plugins.
If you have someone on retainer see if you can get them to clean up the database.
3)
Hosting issues, and things which can be fixed at the hosting tier.
Server load is the big one here. On our slower days you shouldn't worry. Like on Sundays, and even then we have really fast solid state storage drives for your site(s).
One thing to note is if you are using the really-simple-ssl plugin. It used to be fantastic, but recent versions of it are getting slower and slower. Ask us to convert your site for ssl/HTTPS for $25, and we can eliminate that as a possibility.
Our "normal" setting for timeouts is 30 seconds. Should you have an issue or specific request requiring a longer timeout to be applied we do have an experimental setting to extend that time. Extending it does make us more vulnerable to various "Denial of Service" attacks where the bozos keep opening connections to the server to tie up all the resources they can. If we do apply that for a circumstance, we also reserve the right to move it back to 30 seconds should we deem it necessary.