[1.1.0-0ubuntu1~bzr341] Nearly hangs even modern computer while indexing large file

Bug #492287 reported by Dmitry Tantsur
This bug report is a duplicate of:  Bug #376896: Excessive swapping during hash. Edit Remove
10
This bug affects 2 people
Affects Status Importance Assigned to Milestone
LinuxDC++
Incomplete
Undecided
Unassigned

Bug Description

LinuxDC++ indexing is too agressive. When it indexes a large file on my Core2Duo notebook all other program nearly stop to response, sound (Audacious) starts to interrupt. So I've to close LinuxDC++ when I need my notebook if it's indexing something.

I tried to set nice to +15, that helped but not perfectly. Maybe indexing should be less agressive?

Ubuntu 9.10, LinuxDC++ 1.1.0-0ubuntu1~bzr341 (from ppa).

Revision history for this message
Razzloss (razzloss) wrote :

And your kernel version is? This sounds like a duplicate of Bug #376896. And if not there's already an option in Advanced->experts only -> Max hash speed for limiting the hashing speed.

--RZ

Changed in linuxdcpp:
status: New → Incomplete
Revision history for this message
Dmitry Tantsur (divius) wrote :

It's a different case. My kernel is 2.6.31 (Karmic), and I have vm.swappiness = 0.

I'll try to play with this expert option, thank you. But maybe it shouldn't be an "expert-only" option? After all it's hard to break a program via it.

Revision history for this message
Razzloss (razzloss) wrote :

OK, I have no idea if that kernel behaves any differently.

I think that option was (at least in early versions) there in vanilla dc++ also (that's pretty much the reason why it's there in linuxdcpp). Also true that you really can't break anything with that, but most users want their shares hashed as fast as possible, so that makes it a bit of an expert option.

But hashing should not cause your Core2Duo to slow to a crawl unless 1) there's the swapping problem (check with free or top if swapping occurs while hashing), 2) You have a problem with DMA (it isn't enabled) 3) you have a blazing fast disks (which I doubt since you were running on a notebook). Those are the most common culprits that I could think of, but of course there's always a possibility that there's some bug hiding somewhere and in that case we'd need to figure out how to reproduce it (or a patch fixing it is always welcome ;)

--RZ

Revision history for this message
Dmitry Tantsur (divius) wrote :

I'll try to investigate the case.

To post a comment you must log in.
This report contains Public information  
Everyone can see this information.

Other bug subscribers

Remote bug watches

Bug watches keep track of this bug in other bug trackers.