Hartmut Figge wrote:
User:
An update to the reoccuring stalling mystery of SM 2.0 that I posted in detail
earlier. After many hours of close observation of the CPU load on the individual
CPU cores on the multi-core CPU I am running on, using both the AMD Power
Monitor (which displayed individual core load and dynamic core clock speed), as
well as the Windows-XP Task Manager PERFORMANCE tab CPU activity graphs. I have
noted that whenever the SM 2.0 stalling occurs every few minutes, the CPU load /
utilization on one of the 3 cores on the processor reaches nearly full load 100%
for a number of seconds (often 10 to 20 seconds), during which time the entire
SM 2.0 suite and all its windows will freeze / stall temporarily. This
observation has been consistent all this time.
That reminds me on<4ae69e4e.60...@hfigge.myfqdn.de> in
de.comm.software.mozilla.nightly-builds. For me the increasing load
occurs every time when the newsservers are updated due to 'Check for new
messages every [ ] minutes'. Or at the start of MailNews.
First i had posted about this issue several weeks ago in
<4abf7628.4060...@hfigge.myfqdn.de>, also dcsm.n-b. And later i thought
that some numbers would be nice. *g*
http://www.triffids.de/pub/sm/2.0/load/smlog.txt
In the lines starting with 'top' the relative time of the measurement
can be seen. Every second. Interesting is the number below %CPU.
Haven't investigated further. But you are not alone, it seems. ;)
Hartmut
Thanks very much for your reply. Finally - an explanation for the first time on
this mystery that makes a lot of sense. I did already notice that when loading
newsgroups or downloading large number of newgroup headers, or when removing
expired newsgroup articles, that considerable CPU load results. It just never
occured to me that the auto check newsgroup messages would cause this sort of
serious load on the entire suite to the point of stalling.
I've since inspected the auto check server settings for the several newsgroup
servers I have set up with SM 2.0 (all newsgroups of which were successfully
migrated manually from SM 1.X) and found that there were 2 news servers (the
Mozilla server included) that had auto check message function on. From what I
remember, I had these preferences on in SM 1.X as well but never appeared to
encounter the CPU load issue in SM 1.X and didn't think much about it when
migrating to SM 2.0.
However, after disabling the auto check message function, virtually all of the
stalling appeared to have disappeared. I have only used SM 2.0 for about an hour
since this making this adjustment, but overall everything is working a lot more
smoothly though there is still the occasional minor stuttering of video playback
(I tried only Flash FLV playback in web pages so far), and Flash FLV video
playback within a web page appears to still generate more CPU load compared to
SM 1.X, but the demands are still well within available CPU power from my
multi-core CPU.
One thing I have never been too clear about the newsgroup function on the
Mozilla suite / Seamonkey - if I have a newsgroup server collapsed in the news
folder pane, then the newsgroup server is NOT accessed UNTIL I either highlight
the news server or open it and highlight one of the newsgroups hosted on the
server, is that correct? Normally, I do not access newsgroups much and would
like to keep newsgroup functions CPU loads to zero or minimal levels. If you or
anyone can clarify this point about the newgroup function, I would appreciate it.
_______________________________________________
support-seamonkey mailing list
support-seamonkey@lists.mozilla.org
https://lists.mozilla.org/listinfo/support-seamonkey
