Stefan Küng wrote:
> If you can ever come up with a recipe on how to get the cache to use
> all CPU I'd really appreciate it.
>
> Stefan
>
Point taken about the Subject line.
This isn't so much a recipe because there is so much going on behind the
scenes, perhaps after internal delays, so the correlation of action and
re-action is difficult.
It's like trying to find out who farted at the dinner table. Nothing
owns up.
As far as I can ascertain it only occurs once per install of revs > 4939
(On XP and Win2K).
Once it has occurred and been killed there are no further occurrences
It does not re-occur upon re-boot.
Significant changes can be made to the WCs and there is no re-occurrence.
It does not re-occur upon re-boot on a later date.
Reverting back to rev 4949 from 4954 caused the CPU > 90% problem.
Loading and exercising rev 4961 (after unloading rev 4949) re-introduces
the problem.
Re-installing the same rev (4961) and browsing, updating etc does not
cause a re-occurrence.
Repeated process of un-install.
Install 4949
Browse,
On each change of drive or folder there is a momentary few% CPU use by
TSVNCache.
(low CPU expected, i.e. not the consumption associated with a crawl)
It's not until an Update (albeit on an already updated WC file:/// on
local drive) that the cache starts to crawl.
Changing folders mid crawl (which takes several minutes on XP box) send
CPU back to 0% before crawl re-commenced.
Whilst crawling CPU shot to 99%.
Repeated process of un-install.
Install 4961
Browsed to drive with revisioned folders.
A momentary few% CPU use by TSVNCache. (low CPU expected, i.e. not the
consumption associated with a crawl)
Waited < 20 secs CPU shot to 99%
Killed.
Does not occur again unless we re-cycle a previous nightly.
This seems to be the pattern on both the XP and Win2K boxes.
We are in effect, going around the table to see who is not smiling.
Whoever (or in this case whatever) is left must be the culprit.
So, what is different about a first ever start of a changed rev vs any
other form of start.?
Peter
Apologies about the graphic analogy if your offended, maybe it should
just be labeled "process of elimination".
---------------------------------------------------------------------
To unsubscribe, e-mail: dev-unsubscribe@tortoisesvn.tigris.org
For additional commands, e-mail: dev-help@tortoisesvn.tigris.org
Received on Fri Nov 18 02:06:21 2005