Re: AW: Status of FailSafe on RH7.2 with GLIBC 2.2.4

the cdbd is multithreaded and on linux, every thread looks as a process
if you make ps... 

the best example is the java runtime.. starting a java process with
native multithreading starts for every thread a new "clone" process due
to the different thread implementation in linux.

so every  failsafe process using the cdbd process, starts a new rpc
connection. every rpc connection gets a new thread...  you can check it,
if you start a mgt-client... check the amount of cdbd process...


oliver


ps: if you have a short look on my patches, you will see, that only the
destroy of the rpc handle is pushed on a stack for later releasing by
the main thread.. all other thread handling stuff is the same like
before in 1.0.3 ... 





> Over the weekend I've set up failsafe on a redhat 7.2 box, using the oss cvs version + suse 1.0.3 patches(without init-script changes) + Oliver Jehles glibc patches.
> 
> Configuration summary:
> 
> Mon Apr 22 13:16:43 CEST 2002
> Resource_group db:
>         State: Online
>         Error: No error
>         Owner: webc1
>         Failover Policy: default_webc1
>         Resources:
>                 /dev/nb0        (type: drbd)
>                 193.154.7.9     (type: IP_address)
>                 /var/dbase      (type: Filesystem)
>                 pgsqlctl        (type: postgres)
> Resource_group web:
>         State: Online
>         Error: No error
>         Owner: webc2
>         Failover Policy: default_webc2
>         Resources:
>                 193.154.7.10    (type: IP_address)
>                 193.154.7.8     (type: IP_address)
>                 /usr/www        (type: Filesystem)
>                 /dev/nb1        (type: drbd)
>                 apache  (type: Apache)
> Seems to be working, but a couple of questions have come up:
> 
> * cdbd: the number of cdbd processess running on my nodes is rather frightening: 32 jobs on node 1 and 43 on node 2.  Should I consider this normal or is this unexpected?
> 
> * what's the correct way of having ha services compe up automaticaly after reboot? basic cluster functions are started using the fs_cluster service script, but I'm not sure on how/when to autostart failsafe.
> 
> Thanks, Martin
> _______________________________________________
> LinuxFailSafe mailing list
> [email protected]
> http://lists.community.tummy.com/mailman/listinfo/linuxfailsafe