Strange occasional "the site not available" error

Events happening in the community are now at Drupal community events on www.drupal.org.
ajayg's picture

I am getting a strange "The site is currently not available due to technical problems" error but only occasionaly on a test server. There is no traffic on the server other than I manually testing. The strange thing is

  1. Error happens only at first access of the domain after a long time
  2. The immediate refresh works fine and sites shows again.
  3. It shows up for several minutes (possibly hours) fine.
  4. After some time (I am trying to nail this down) if you try again it suddenly shown the same error message.

The setup is 1 webserver and 1 database server. The server logs and other performance graphs do not show any unusual traffic when I am testing. In fact my testing causes very negligible load.

I am suspecting some kind of connection timeout happening between webserver and db server since I have heard the same message showing when there is too much load or connection issues on the server. But I don't get why I get this message first time and it won't try to connect very first time.

Comments

if Drupal have problems with

Alexander Kosarev's picture

if Drupal have problems with database (permissions, table miss, including timeouts) you will see notice about this. Error message showed and tell you want is happened.

If you have only "The site is currently not available due to technical problems", this can mean anything, Did you try to clear cookies and access again?
Do you use module load performance? (boost, cache router or another) - it is possible to be session initialization problem.

if there is a permission

ajayg's picture

if there is a permission problem I would expect the error everytime. Not once a day. In fact several hours have gone since I wrote this but I haven't seen the issue yet today. If there is a cookie issue then it would show everytime (or atleast with more frequency)

Yes I have both boost and cache router enabled. However I have multiple sites on the same server and the problem shows even on sites where those two modules are disabled. I also have serverwide APC.

(But I have APC on a different setup too and that setup does not show this issue.)

the problem goes away (for several hours) by just reloading the page in browser.

you have wrote

Alexander Kosarev's picture

you have wrote that you site have not load, can you disable boost and cache router and try to reproduce problem?

Look like this problem isn't 'database' related, I think something happened with session:
- with first site visit session initialized and you see problem
- next visits (including refresh) you don't see problem
- after few hours(session destroyed) problem come back

can you monitor your session via session table in Drupal database?

I am also thinking this

ajayg's picture

I am also thinking this related to session. I am trying to figure out how to monitor the session. The problem is I don't know yet after how long the problem shows up so don't know how long should I monitor. I continue to investigate.

problem went away on a busy

ajayg's picture

problem went away on a busy site.
I since started using this server on production and have never seen the issue. I am suspecting there is some kind of idle timeout property kicking somewhere which was disconnecting connection with DB. Since there is almost stead traffic this problem seems to be gone.

I have it on my website too,

asinnema's picture

I have it on my website too, no boost enabled, no cache enabled. I tried to enable cache but the problem persists.

Subscribing! If anyone figures out what to do - except make sure my website is bombarded with visitors ;) I'd love to hear it.

High performance

Group notifications

This group offers an RSS feed. Or subscribe to these personalized, sitewide feeds: