MRatWork Forum by Mustafa Ramadhan
Sawo Project - Kloxo-MR Discussions => Kloxo-MR Bugs and Requests => Topic started by: Spacedust on 2014-09-13, 15:13:43
-
Nobody sees this ? chris ?
It all started after KloxoMR 7.0b with new configs
You just add domain:
- pages are stalled
- then pages instantly shows nginx error message
- then nothing is working at all apache+nginx died - no such server message from the browser
- then after a minute apache+nginx is restarted by watchdog and everything is working fine again
[Sat Sep 13 01:33:22 2014] [notice] FastCGI: process manager initialized (pid 31962)
[Sat Sep 13 14:06:16 2014] [notice] FastCGI: process manager initialized (pid 9202)
[Sat Sep 13 15:07:18 2014] [alert] (4)Interrupted system call: FastCGI: read() from pipe failed (0)
[Sat Sep 13 15:07:18 2014] [alert] (4)Interrupted system call: FastCGI: the PM is shutting down, Apache seems to have disappeared - bye
[Sat Sep 13 15:07:39 2014] [notice] FastCGI: process manager initialized (pid 19808)
[Sat Sep 13 15:08:16 2014] [notice] FastCGI: process manager initialized (pid 21851)
[Sat Sep 13 15:10:20 2014] [alert] (2)No such file or directory: FastCGI: read() from pipe failed (0)
[Sat Sep 13 15:10:20 2014] [alert] (2)No such file or directory: FastCGI: the PM is shutting down, Apache seems to have disappeared - bye
[Sat Sep 13 15:10:40 2014] [notice] FastCGI: process manager initialized (pid 28093)
[Sat Sep 13 15:11:17 2014] [notice] FastCGI: process manager initialized (pid 29858)
-
See how it looks like !
It says no such file or directory in one place. But which file ???
(http://files.tinypic.pl/i/00575/5zo7uuxxlqkg.jpg)
-
There are some nasty bugs - Mustafa fix this immediately !!!
This is the first one:
FCGIWrapper /home/kloxo/client/php5.fcgi .php
but Apache+nginx still dying even with only one domain.
-
Still not fixed !
[root@c44-49 /]# tail -f /var/log/httpd/error_log
[Mon Sep 15 01:20:23 2014] [warn] RSA server certificate wildcard CommonName (CN) `*.lxlabs.com' does NOT match server name!?
[Mon Sep 15 01:20:23 2014] [warn] RSA server certificate wildcard CommonName (CN) `*.lxlabs.com' does NOT match server name!?
[Mon Sep 15 01:20:23 2014] [notice] Digest: generating secret for digest authentication ...
[Mon Sep 15 01:20:23 2014] [notice] Digest: done
[Mon Sep 15 01:20:23 2014] [notice] FastCGI: process manager initialized (pid 1984)
[Mon Sep 15 01:20:23 2014] [warn] RSA server certificate wildcard CommonName (CN) `*.lxlabs.com' does NOT match server name!?
[Mon Sep 15 01:20:23 2014] [warn] RSA server certificate wildcard CommonName (CN) `*.lxlabs.com' does NOT match server name!?
[Mon Sep 15 01:20:23 2014] [warn] RSA server certificate wildcard CommonName (CN) `*.lxlabs.com' does NOT match server name!?
[Mon Sep 15 01:20:23 2014] [warn] RSA server certificate wildcard CommonName (CN) `*.lxlabs.com' does NOT match server name!?
[Mon Sep 15 01:20:23 2014] [notice] Apache/2.2.29 (Unix) DAV/2 mod_fastcgi/2.4.7 mod_rpaf/0.6 mod_ssl/2.2.29 OpenSSL/1.0.1e-fips mod_perl/2.0.4 Perl/v5.10.1 configured -- resuming normal operations
[Mon Sep 15 01:20:57 2014] [notice] caught SIGTERM, shutting down
[Mon Sep 15 01:20:57 2014] [alert] (2)No such file or directory: FastCGI: read() from pipe failed (0)
[Mon Sep 15 01:20:57 2014] [alert] (2)No such file or directory: FastCGI: the PM is shutting down, Apache seems to have disappeared - bye
^C
[root@c44-49 /]# yum upgrade
Loaded plugins: fastestmirror, priorities, replace
Loading mirror speeds from cached hostfile
* base: www.gtlib.gatech.edu
* elrepo: ftp.nluug.nl
* epel: mirror.its.dal.ca
* extras: www.gtlib.gatech.edu
* mratwork-epel: mirror.its.dal.ca
* mratwork-ius-stable: mirror.amsiohosting.net
* rpmforge: mirror.nl.leaseweb.net
* updates: www.gtlib.gatech.edu
Setting up Upgrade Process
No Packages marked for Update
-
Happens on mod_ruid2 and suphp too.
-
Happens on Apache only configuration too.
-
Happens on nginx only configuration too:
2014/09/15 01:32:14 [crit] 12186#0: *3 open() "/home/httpd/tester.xxx.pl/stats/tester.xxx.pl-custom_log" failed (13: Permission denied) while logging request, client: 62.21.44.49, server: tester.xxx.pl, request: "GET /images/kloxo-mr.png HTTP/1.1", host: "www.tester.xxx.pl", referrer: "http://www.tester.xxx.pl/"
-
Still not fixed even on Kloxo-MR 7.0.0.b-2014091602 with /script/cleanup. All pages are down till Kloxo executes magic commands:
21:03 Sep/16/2014: Closed Exec /usr/local/lxlabs/kloxo/cexe/closeinput '/etc/init.d/httpd graceful' >/dev/null 2>&1 &
21:03 Sep/16/2014: Closed Exec /usr/local/lxlabs/kloxo/cexe/closeinput '/etc/init.d/nginx restart' >/dev/null 2>&1 &
-
Every delete/add domains, kloxo will be execute restart for services in queue (execute with createRestartFile).
-
Every delete/add domains, kloxo will be execute restart for services in queue (execute with createRestartFile).
I know, but nginx+apache is dying before Kloxo executes /etc/init.d/nginx reload +/etc/init.d/httpd graceful
so you have about 20-60 seconds downtime on each domain added/removed...
It happens even on clean system with just 1 domain !
-
Interesting thing - if I use Kloxo-MR 6.5.1.b-2014070704 folder from backup then it all works again, so our configs are fine, the problem is inside Kloxo code. What was changed ???
-
Happens on Kloxo-MR 7.0.0.b-2014082703 too. I need to test older versions to see where this bug was introduced.
-
Suprise - Kloxo-MR 7.0.0.b-2014082402 works perfectly - so there must be a bug introduced from this version.
kloxomr7-7.0.0.b-2014082701.mr.noarch - is not working ok
-
I've tested latest configs with version Kloxo-MR 7.0.0.b-2014082402 and all works fine.
Something is inside Kloxo source code which is causing this ! I suspect httpdocs directory.
-
The problem is inside /usr/local/lxlabs/kloxo/httpdocs/driver/web directory.
Now we have to find which file is causing this !
-
Every process to add/delete/update will be execute 'sh /script/fixweb --domain=<domain>'.
Because at previous code, except '--target=defaults', will be delete all domains configs and it's make trouble for '--domain='. So, in latest update, this issue already fixed. Read fixweb.php.
-
OK. I'm updating and testing.
-
Just updated to Kloxo-MR 7.0.0.b-2014091704 and Apache+nginx is still dying so nothing was fixed.
-
It seems /etc/init.d/nginx reload is not responding after adding domain:
[root@c44-49 ~]# /etc/init.d/nginx reload
[root@c44-49 ~]#
because it's dead (while it should be running):
[root@c44-49 ~]# ps aux | grep nginx
root 29775 0.0 0.0 103260 892 pts/1 S+ 13:42 0:00 grep nginx
so we have to restart it:
[root@c44-49 ~]# /etc/init.d/nginx restart
Zatrzymywanie nginx: [NIEUDANE]
Uruchamianie nginx: [ OK ]
-
Need waiting 5-10 secs for restart web services.
-
I know, but it's dying completely, with old drivers everything is working great:
See the video: http://galeriaportali.pl/kloxomr.wmv
-
Still not fixed in Kloxo-MR 7.0.0.b-2014092002 - I have to restart nginx manually, watchdog does not restart it.
-
Try copy /usr/local/lxlabs/kloxo/sbin/kloxo.php to custom.kloxo.php and modified 'timed_exec' value.
-
It does not help. Mustafa let me explain this one more time.
Old driver allowed nginx to just reload graceful after each domain was added/removed. No downtime !
New driver kills nginx so we have about 60-120 seconds downtime till it will be restarted - kloxo it trying to reload dead nginx but that is not possible !
Restore the old driver it was working much better.
-
What's nginx version better according your opinion.
Before centalt repo down, Kloxo-MR use nginx from centalt but after that use from nginx.org directly.
-
What's nginx version better according your opinion.
Before centalt repo down, Kloxo-MR use nginx from centalt but after that use from nginx.org directly.
nginx version does not matter, because it works fine. I'm using the latest upload with old webdrivers for Kloxo and all works fine. Kloxo is reloading nginx and gracefulling apache so there is no downtime when adding/deleting domains (nginx is displaying a copy of page from it's cache while apache is not ready).
-
Update your Kloxo-MR 7.0.0.
-
It works again - thanks !!!
-
Still crashing when removing customers with many domains, FTP accounts etc.
-
Still crashing when removing customers with many domains, FTP accounts etc.
Try update.
-
I know what's the problem ! Kloxo-MR since version 7.0b does not reload php-fpm so fast after reloading gracefulling apache.
I'm on Kloxo-MR 7.0.0.b-2014092203 now