The last 5 days our server crashes every day at some time between 6:00 and 7:00. Only ping seems to work (must check this to be sure) - signalling, that the kernel still lives. The crash (or the restart) is so hard, that the software-raid was degraded after one crash und I had to rebuild it. As the exakt time of the crash changes, no cron-job should be the cause. The server ist a root-server located at the provider, so I cannot put my hands on it.

I can't believe, that this has to do anything with Limesurvey, but - the only change in this period was the start of our first survey with a load of some 100 to 150 (small) surveys filled in every day (at daylight, not in the morning!). Still can't believe it, but if anyone had a similar experience, I would be glad to know it.

Thanks and kind regards
Erich

A crash every day around the same time sounds very unlikely to be from Limesurvey. I never had a server crash caused by Limesurvey anyway. Because even if it gets overloaded it should not really crash I think.

When you say "100 to 150 (small) surveys", what does "surveys" stand for? Filled in questionnaires / responses or do you have 100-150 different surveys running at the same time and a lot of people are filling all these surveys? But even then, it doesn't really make sense that the server crashes around the same time, especially as at this time of the day response rate shouldn't be too high.

What software the server is running? I had some problems in the early morning with some server services stopping on a VPS every day (I got notificaction through Cpanel). Reason was the backup the provider was running during this time...

To answer your questions: I meant "100 to 150 filled in questionaires" - so less, the statistics don't even recognize the load. The server ist called "root-server", meaning that we are the one and only user and it runs Centos. There ist no support from the provider, even no backup. The only applications on that server are Apache, a small homepage with small traffic and Limesurvey. The server runs without problems since half a year, running Limesurvey as a test. Practical usage started about two weeks ago and some days later the problems began.

Our customer plans to run up to 500 surveys with each some 200 participants per year as a service for his customers. Not too much, I think. His personal manages Limesurvey via a UI written by us.

In the meantime I've got one more information: no crash on saturday and sunday morning. So the cause probably is something happening only on workdays either insider the server (an application) or outside the server, in the environment. My favorites are

- still Limesurvey or our extension written in PHP and using the Remot Control API
- something the cleaning personal does or
- something happening with the power system. e.g. a power glitch caused by an electrical door or else or
- a change in temperature in combination with a contact problem.

As our provider is not willing to give much information about the environment, I asked here, just to be sure. Now knowing that Limesurvey probably should not be the cause, my favorite ist to look once again through our simple PHP-extension (together some 300 statements), though I never had PHP crashing a Server. Thank you for your interest and any more idea you may have.

Kind regards,
Erich

This sounds as a setup that is weak for the purpose you want to use it for.

I would personally never use a server without replication to another server and I would not trust a provider that does not make backups and does not give information. You are planning data acquisition, in that process "trust" is an important word.

Just my 2 cents,

Tammo

What do you mean with "weak setup"? Regarding what you say about the need of replication and backup, I fully agree. We are doing both.

Our customer is very sensitive in the area of data security and asked us, to use what our provider calls "root server". This is a piece oft hardware connected to the internet and containing only an installer. You have to install even the operating system and the rest you need. Only you as the "owner" have access to the server, so you have to manage the whole server and backup also. That means, you're relative alone in case of problems. If the server crashes, the only thing we can do remote, is a cold restart.

Kind regards,
Erich

Hi Erich,

I was referring to your statement "There ist no support from the provider, even no backup." I thought that indicated that you operate without help or backup, something innocent users might do. I did not know that you have it all arranged...

For us security is also always very important, I have seen organisations that say that, but act otherwise. We also use dedicated servers where needed.

Greetings from the Netherlands,

Tammo

No, Limesurvey was not the cause of our problems. It was our provider (one of the biggest in Gemany!), twiddling with the internet connection of the server. When starting "serious" work we noticed the next day, that we couldn't reach the server. After some minutes we thought, the server had crashed and made a remote restart (hard restart). Next day, same situation. This time, the server probably was doing some cron job and the reset occured in a moment when the Raid-software tried to write some data to the mirror disk. After the restart the raid-array was reported as "degraded", increasing F.U.D. (fear, uncertainty, doubt) for us.

To make it short: the server runs, the Raid-array is healthy and the provider has "never heard of any interruption".

Thanks for all comments and kind regards
Erich

Thank you for the feedback!

The big provider in Nuremberg or the one in Berlin?

tammo wrote: I would personally never use a server without replication to another server

What kind of replication are you talking about? Would be interesting to know where you got Limesurvey running on with replication flawlessly.

@jelo: On production server 1 we do a MySQL dump and a rsync of the files to production server 2 every night. On top of that a cyclic 7-days MySQL dump is created on backup server 1.

On production server 2 we have MySQL replication running constantly to 2 other servers. Every night Directadmin on that server creates a local backup of the whole server (users, files, MySQL). That full backup (7GB) is rsynced to backup server 2 at the end of the night. Production server 2 is also part of the 7-day cyclic MySQL dump on backup server 1.

Our provider (CloudVPS) has a feature to clone one virtual server to another. Every weekend we clone production server 2 to backup server 2. This is the only backup we do manually. All other backups are invoked with cron jobs.

@jelo: JanE is my business partner; together we are Tools for Research...