v4 won't (re) start if _.vsm preexists -- starts fine if deleted 1st
grantksupport at operamail.com
grantksupport at operamail.com
Mon May 5 16:13:28 CEST 2014
Hi,
Still working on my v3->v4 upgrade.
I've noticed that while v3 stopped/started without any problems, my v4
instance starts -- but only (?) if there's no previous "_.vsm" lying
around.
Here's what I'm seeing:
systemctl start varnish-custom.service
-> RUNS/OK
ls -al /usr/local/var/varnish/TEST/
total 82M
drwxrwxrwt 2 root root 80 May 5 06:55 ./
drwxrwxr-x+ 3 grant users 4.0K May 5 06:05 ../
-rwxr-xr-x 1 nobody nogroup 77K May 5 06:55 vcl.vLB_IZSL.so*
-rw-r--r-- 1 root root 81M May 5 06:55 _.vsm
systemctl stop varnish-custom.service
ls -al /usr/local/var/varnish/TEST/
total 120M
drwxrwxrwt 2 root root 80 May 5 06:56 ./
drwxrwxr-x+ 3 grant users 4.0K May 5 06:05 ../
-rw-r--r-- 1 root root 81M May 5 06:55 _.vsm
-rw-r--r-- 1 root root 39M May 5 06:56 _.vsm.13175
systemctl start varnish-custom.service
Job for varnish-custom.service failed. See 'systemctl status
varnish-custom.service' and 'journalctl -xn' for details.
systemctl stop varnish-custom.service
systemctl start varnish-custom.service
Job for varnish-custom.service failed. See 'systemctl status
varnish-custom.service' and 'journalctl -xn' for details.
systemctl stop varnish-custom.service
ls -al /usr/local/var/varnish/TEST/
total 120M
drwxrwxrwt 2 root root 80 May 5 06:56 ./
drwxrwxr-x+ 3 grant users 4.0K May 5 06:05 ../
-rw-r--r-- 1 root root 81M May 5 06:55 _.vsm
-rw-r--r-- 1 root root 39M May 5 06:56 _.vsm.13175
rm -f /usr/local/var/varnish/TEST/_.vsm*
ls -al /usr/local/var/varnish/TEST/
total 4.0K
drwxrwxrwt 2 root root 40 May 5 07:01 ./
drwxrwxr-x+ 3 grant users 4.0K May 5 06:05 ../
systemctl start varnish-custom.service
-> RUNS/OK
ls -al /usr/local/var/varnish/TEST/
total 82M
drwxrwxrwt 2 root root 80 May 5 07:03 ./
drwxrwxr-x+ 3 grant users 4.0K May 5 06:05 ../
-rwxr-xr-x 1 nobody nogroup 77K May 5 07:03 vcl.qH7WsF0n.so*
-rw-r--r-- 1 root root 81M May 5 07:03 _.vsm
Reading here,
https://www.varnish-cache.org/docs/trunk/reference/vsm.html, iiuc (?)
the vsm should be cleared on stop?
I can reproduce this ^^^ on two different machines, so if it's something
I'm (not) doing, I'm at least consistent.
Grant
More information about the varnish-misc
mailing list