Hey All,

I ran the problem vault earlier last night so that I could watch the 
memory while it was running.  I dumped the memory on both the client 
(the problem machine) and the dirvish backup server.  The results can be 
found below.  On the client we can see there's still 38MB free.  That, I 
assume is not the problem.  The server shows 5MB of free memory.  That 
may be the problem, but I'm not sure.  Can somebody please weigh in?

Problematic Client During Dirvish Run:
MemTotal:      1026120 kB
MemFree:         38516 kB
Buffers:        216588 kB
Cached:         340256 kB
SwapCached:          8 kB
Active:         414076 kB
Inactive:       355944 kB
HighTotal:      121376 kB
HighFree:          204 kB
LowTotal:       904744 kB
LowFree:         38312 kB
SwapTotal:     3903672 kB
SwapFree:      3903256 kB
Dirty:            2780 kB
Writeback:          92 kB
Mapped:         271948 kB
Slab:           200884 kB
CommitLimit:   4416732 kB
Committed_AS:   296916 kB
PageTables:       3304 kB
VmallocTotal:   118776 kB
VmallocUsed:      6300 kB
VmallocChunk:   111972 kB

Problematic Client Without Dirvish Running:
MemTotal:      1026120 kB
MemFree:         64520 kB
Buffers:        200852 kB
Cached:         498700 kB
SwapCached:         32 kB
Active:         557828 kB
Inactive:       278900 kB
HighTotal:      121376 kB
HighFree:          144 kB
LowTotal:       904744 kB
LowFree:         64376 kB
SwapTotal:     3903672 kB
SwapFree:      3903228 kB
Dirty:              28 kB
Writeback:           0 kB
Mapped:         208904 kB
Slab:           108288 kB
CommitLimit:   4416732 kB
Committed_AS:   224536 kB
PageTables:       3284 kB
VmallocTotal:   118776 kB
VmallocUsed:      6300 kB
VmallocChunk:   111972 kB


Here are the reports from the server:
------------------------------
While the backup is running:
------------------------------
MemTotal:       515612 kB
MemFree:          5492 kB
Buffers:         21264 kB
Cached:         327824 kB
SwapCached:          0 kB
Active:         165856 kB
Inactive:       319344 kB
HighTotal:           0 kB
HighFree:            0 kB
LowTotal:       515612 kB
LowFree:          5492 kB
SwapTotal:     3903672 kB
SwapFree:      3903500 kB
Dirty:              68 kB
Writeback:           0 kB
AnonPages:      136156 kB
Mapped:           9656 kB
Slab:            16968 kB
SReclaimable:    10216 kB
SUnreclaim:       6752 kB
PageTables:       1392 kB
NFS_Unstable:        0 kB
Bounce:              0 kB
CommitLimit:   4161476 kB
Committed_AS:   259660 kB
VmallocTotal:   511992 kB
VmallocUsed:      4716 kB
VmallocChunk:   506992 kB

-------------------------------
While the backup is NOT running
-------------------------------
MemTotal:       515612 kB
MemFree:         11316 kB
Buffers:         16804 kB
Cached:         426520 kB
SwapCached:          0 kB
Active:          50764 kB
Inactive:       428252 kB
HighTotal:           0 kB
HighFree:            0 kB
LowTotal:       515612 kB
LowFree:         11316 kB
SwapTotal:     3903672 kB
SwapFree:      3903488 kB
Dirty:               4 kB
Writeback:           0 kB
AnonPages:       35700 kB
Mapped:           8660 kB
Slab:            17668 kB
SReclaimable:    10588 kB
SUnreclaim:       7080 kB
PageTables:        860 kB
NFS_Unstable:        0 kB
Bounce:              0 kB
CommitLimit:   4161476 kB
Committed_AS:    61604 kB
VmallocTotal:   511992 kB
VmallocUsed:      4716 kB
VmallocChunk:   506992 kB


[EMAIL PROTECTED] wrote:
> If it worked well for a while and is now seemingly taking
> exponentially long to finish, I'd make sure you're not swapping
> excessively.  A directory of mail probably has a much larger ratio of
> files to bytes than other places in your filesystem, so just saying
> "there's not much data there" may misdiagnose the problem.
> 
> Since you're (presumably) asking rsync to transfer a large number of
> files, rsync must hold information about all of them in its address
> space while it runs, and it's possible that you've crossed some magic
> threshold (as the number of users has grown) that is causing rsync not
> to fit in physical memory after it's built that data structure.  If
> that's true, you're going to run very, very slowly as all those pages
> get thrashed onto and off of the disk again.
> 
> [And if you have dirvish passing -H to rsync, this is even more likely
> (especially with older rsync versions)---and isn't fully fixed even in
> the most-recent rsync release (since -H makes it hard for rsync to
> know when it can forget about already-transferred directories if
> there's a possibility of a hardlink in there somewhere).]
> _______________________________________________
> Dirvish mailing list
> Dirvish@dirvish.org
> http://www.dirvish.org/mailman/listinfo/dirvish
> 
> 


_______________________________________________
Dirvish mailing list
Dirvish@dirvish.org
http://www.dirvish.org/mailman/listinfo/dirvish

Reply via email to