We have a Drupal 4.5 site running on a Pentium 4 with 1GB RAM (shared with other sites: one Drupal 4.5, some WordPress, some MT) and it might be that Drupal has been responsible for hanging it several times. The server gets collapsed and the only solution is to restart it - the hard way.
(Ruby On Rails was the suspect so we move it to another server, but now it has happened again)
The site < http://www.popmadrid.com > is a music community/ecommerce site [1], with two vocabularies (bands and labels) with +700 (seven hundren) terms each (being this huge vocabs the new suspects...). And not much traffic: 1200 users daily and 5000 page views.
We activated the MySQL logs (general and slow) since last time and this morning the server crashed again. So we have fresh material to look at :)
But it shows nothing strange :( (or at least something I can't get a clue from)
At 7:55 this morning mysqlslow starts to get fillep up with slow querys from the DB Drupal is running on (and some querys from another apps). They are normal Drupal querys from core modules - The first query start at Query_time: 6; with many querys each second. Gradually (for 3-4 minutes) querys starts taking more time, 10-20, until some peaks of 120-130 query time and then query times between 40/60 and 100/150, with some very huge UPDATE cache SET data in the middle, until 8:25.
At 8:24 with get query times of 400 up to 1200 at 8:35, the time at which messagges log stopped loggin anything. mysqlslow kept logging querys up to 2000 query time until we restarted.
Some hint of what might be happening?
ciao álvaro
Hi alvaro! that website looks great!
That is totally weird! .. one of my main reason to moving to drupal a month ago was that some perl CMS were killing the server and not the php ones.
The first thing you do is enable throttle which handles the auto-throttling mechanism, to control site congestion. just to make sure that we are not querying more than we have to.
while you check your logs I suggest what I was doing when I had those perl based CMS:
create a cron like:
0 */6 * * * /etc/init.d/httpd restart
This will restart your server every 6 hours. So while try to find the problem server cleans it self.
Can you post here your error logs of apache when dies?
BTW! what are you using to show all the content on the 1st page? summary module? (this could be killing it if is doing a lot of querys wrong)
Que tengas un buen dia! y suerte con los logs!
On 5/26/05, Álvaro Ortiz alvaro.ortiz@the-cocktail.com wrote:
We have a Drupal 4.5 site running on a Pentium 4 with 1GB RAM (shared with other sites: one Drupal 4.5, some WordPress, some MT) and it might be that Drupal has been responsible for hanging it several times. The server gets collapsed and the only solution is to restart it - the hard way.
(Ruby On Rails was the suspect so we move it to another server, but now it has happened again)
The site < http://www.popmadrid.com > is a music community/ecommerce site [1], with two vocabularies (bands and labels) with +700 (seven hundren) terms each (being this huge vocabs the new suspects...). And not much traffic: 1200 users daily and 5000 page views.
We activated the MySQL logs (general and slow) since last time and this morning the server crashed again. So we have fresh material to look at :)
But it shows nothing strange :( (or at least something I can't get a clue from)
At 7:55 this morning mysqlslow starts to get fillep up with slow querys from the DB Drupal is running on (and some querys from another apps). They are normal Drupal querys from core modules - The first query start at Query_time: 6; with many querys each second. Gradually (for 3-4 minutes) querys starts taking more time, 10-20, until some peaks of 120-130 query time and then query times between 40/60 and 100/150, with some very huge UPDATE cache SET data in the middle, until 8:25.
At 8:24 with get query times of 400 up to 1200 at 8:35, the time at which messagges log stopped loggin anything. mysqlslow kept logging querys up to 2000 query time until we restarted.
Some hint of what might be happening?
ciao álvaro
--
The Cocktail
wwww -> http://the-cocktail.com blog -> http://furilo.com ring -> +34 91 567 06 05
-- [ Drupal support list | http://lists.drupal.org/ ]
On Thu, 26 May 2005, Álvaro Ortiz wrote:
We have a Drupal 4.5 site running on a Pentium 4 with 1GB RAM (shared with other sites: one Drupal 4.5, some WordPress, some MT) and it might be that Drupal has been responsible for hanging it several times. The server gets collapsed and the only solution is to restart it - the hard way.
(Ruby On Rails was the suspect so we move it to another server, but now it has happened again)
The site < http://www.popmadrid.com > is a music community/ecommerce site [1], with two vocabularies (bands and labels) with +700 (seven hundren) terms each (being this huge vocabs the new suspects...). And not much traffic: 1200 users daily and 5000 page views.
We activated the MySQL logs (general and slow) since last time and this morning the server crashed again. So we have fresh material to look at :)
Since you are not telling us which queries get slow, I'll have to ask my chrystal ball.
My chrystal ball says that some queries in Drupal 4.6 indeed were less than optimal. In particular those for the access controll system. Teh chrystal ball recommends to update to Drupal 4.6.
Cheers, Gerhard
My chrystal ball says that some queries in Drupal 4.6 indeed were less than optimal. In particular those for the access controll system. Teh chrystal ball recommends to update to Drupal 4.6.
Since you had your chrystall ball I did not have to say... :)
The first slow queries were like:
SELECT DISTINCT(n.nid), n.sticky, n.created FROM node n INNER JOIN node_access na ON (na.nid = 0 OR na.nid = n.nid) WHERE n.promote = 1 AND n.status = 1 AND na.grant_view = 1 AND CONCAT(na.realm, na.gid) IN ('all0') ORDER BY n.sticky DESC, n.created DESC LIMIT 0, 10;
The upgrade was in my to-do list for some time now, so the time has come.
Thanks, álvaro -- http://www.furilo.com
On Thu, 26 May 2005, [UTF-8] Ãlvaro Ortiz wrote:
My chrystal ball says that some queries in Drupal 4.6 indeed were less than optimal. In particular those for the access controll system. Teh chrystal ball recommends to update to Drupal 4.6.
Since you had your chrystall ball I did not have to say... :)
The first slow queries were like:
SELECT DISTINCT(n.nid), n.sticky, n.created FROM node n INNER JOIN node_access na ON (na.nid = 0 OR na.nid = n.nid)
^^^^^^ ^ ^^^^^^^^^^^^^^^
We found out that this construct does make for slow queries. It usually does not bring down a server, though. It is however possible, that mysql needs a lot of memory to accomodate for this kind of queries as a lot of table rows need to be looked at.
The upgrade was in my to-do list for some time now, so the time has come.
Yep.
Cheers, Gerhard
-
Carlos Rivero -
Gerhard Killesreiter -
Álvaro Ortiz