[vdr] max. latency time to large, bad response from vdr

[Emil Naepflein <Emil.Naepflein@philosys.de>]

Hi to all,

I have sometime latency times of more than 60 s and had to increase the
watchdog to 120 s now to avoid timeouts. As I have a very busy server
with 4 cards and more than 1 TB of disk space with more than 200
recordings currently, this happens especially when cutting recordings,
removing recordings and doing other reorganizations in the filesystems.
It is especially very annoying when calling the recording menu because
here it takes a very long time to get the recording list. 

I have already replaced the algorithm here and got a considerable
speedup ( < 30 s) during heavy load, but it is still annoying to have to
wait more then 10 s for the recording list. I can improve the response
further by caching the values, but before I make this change I would
like to hear your opinions. 

After tests with bonnie I found out that the best compromise regarding
file read/write and file create/remove performance is the ext3
filesystem. I have also used reiserfs and xfs before but there is not
much difference regarding latency. 

System hardware and software:
AthlonXP 2100+
512 MB Memory
6 x 250 GB Maxtor (Raid5)
2 x 80 GB Maxtor (Raid1)
Linux 2.4.19

So, what is your experience regarding interactive performance and what
can you recommend to improve it further?

Regards,
Emil


-- 
Info:
To unsubscribe send a mail to ecartis@linuxtv.org with "unsubscribe vdr" as subject.