Gentoo Forums
Gentoo Forums
Gentoo Forums
Quick Search: in
Samba CIFS – Filetransfer extremly slow
View unanswered posts
View posts from last 24 hours

 
Reply to topic    Gentoo Forums Forum Index Kernel & Hardware
View previous topic :: View next topic  
Author Message
Palatonka
n00b
n00b


Joined: 12 Mar 2014
Posts: 21

PostPosted: Mon Feb 20, 2017 6:49 am    Post subject: Samba CIFS – Filetransfer extremly slow Reply with quote

After switching kernel from 4.4 to 4.9 cifs filetransfers are extremly slow (more than the fifth!).
Unfortunately I can't imagine what this difference causes.
Using windows share mounted with cifs.

Thanks in advance for any help and kind regards.


Last edited by Palatonka on Mon Feb 27, 2017 10:27 am; edited 3 times in total
Back to top
View user's profile Send private message
Jaglover
Watchman
Watchman


Joined: 29 May 2005
Posts: 7196
Location: Saint Amant, Acadiana

PostPosted: Mon Feb 20, 2017 12:20 pm    Post subject: Re: Kernel 4.4 vs. 4.9 – NFS filetransfer extremly slow with Reply with quote

Palatonka wrote:
After switching kernel from 4.4 to 4.9 NFS filetransfers are extremly slow (more than the fifth!).
Unfortunately I can't imagine what this difference causes.
What has changed in kernel 4.9 regarding NFS that could cause this issue?
Using windows share mounted with cifs.

Confusing. What version of NFS? What has CIFS to do with it?
_________________
Please learn how to denote units correctly!
Back to top
View user's profile Send private message
Palatonka
n00b
n00b


Joined: 12 Mar 2014
Posts: 21

PostPosted: Mon Feb 20, 2017 12:58 pm    Post subject: Reply with quote

Here is the mount info:

//SRV-123/backup on /mnt/backupserver type cifs (rw,relatime,vers=1.0,cache=strict,username=backupusr,domain=mydomain,uid=0,noforceuid,gid=0,
noforcegid,addr=10.10.0.11,file_mode=0755,dir_mode=0755,nounix,serverino,mapposix,rsize=61440,wsize=65536,echo_interval=60,actimeo=1,
user=backupusr)

[Moderator edit: broke very long whitespace-free comma-separated list to improve output layout. -Hu]
Back to top
View user's profile Send private message
Ant P.
Watchman
Watchman


Joined: 18 Apr 2009
Posts: 5878

PostPosted: Tue Feb 21, 2017 5:52 pm    Post subject: Reply with quote

If this actually has anything to do with NFS, post the output of `rpcinfo` on the client/server. Otherwise fix the thread title.
Back to top
View user's profile Send private message
axl
l33t
l33t


Joined: 11 Oct 2002
Posts: 637
Location: Romania

PostPosted: Tue Feb 21, 2017 9:00 pm    Post subject: Reply with quote

Palatonka wrote:
Here is the mount info:

//SRV-123/backup on /mnt/backupserver type cifs (rw,relatime,vers=1.0,cache=strict,username=backupusr,domain=mydomain,uid=0,noforceuid,gid=0,
noforcegid,addr=10.10.0.11,file_mode=0755,dir_mode=0755,nounix,serverino,mapposix,rsize=61440,wsize=65536,echo_interval=60,actimeo=1,
user=backupusr)

[Moderator edit: broke very long whitespace-free comma-separated list to improve output layout. -Hu]


this mount info points to samba not nfs. maybe take a look over the logs of those services. just a thought.
Back to top
View user's profile Send private message
threadau
n00b
n00b


Joined: 22 May 2012
Posts: 15

PostPosted: Sun Feb 26, 2017 2:40 am    Post subject: Reply with quote

I've noticed my Samba access has got extremely slow recently as well - but no kernel change, still on 4.4.x

I'm not talking about 20% speed, though. I'm lucky to get 50 KB/s.
Back to top
View user's profile Send private message
threadau
n00b
n00b


Joined: 22 May 2012
Posts: 15

PostPosted: Sun Feb 26, 2017 5:41 am    Post subject: Reply with quote

As an update...I myself updated to the latest stable kernel (4.9.6), no change.

Don't want to thread-jack, so can start a new one. But my other observations is that I can transfer small files fine (i.e. <100 kB), but larger ones (smallest tried was 30 MB) panic and stop after a couple of MB transferred.
Back to top
View user's profile Send private message
ct85711
Veteran
Veteran


Joined: 27 Sep 2005
Posts: 1701

PostPosted: Sun Feb 26, 2017 7:37 am    Post subject: Reply with quote

This is sounding like the 2 of you may want to go through all changes done on both systems from the time that it was working correctly sense (this does mean it can include a couple months or more of changes). I am assuming you control/manage the server side too, if not just one less system to check (but doesn't mean it's removed from the equation, just makes it an unknown).

Note, this is a good practice to do when troubleshooting an issue (always starting from last known good time). Checking logs is also a good habit to do, even when you don't have an issue. (yes, there is a log of packages installed in the log directory, I'll let you find them among others that you should already be familiar with).
Back to top
View user's profile Send private message
threadau
n00b
n00b


Joined: 22 May 2012
Posts: 15

PostPosted: Mon Feb 27, 2017 10:38 am    Post subject: Reply with quote

Thanks ct85711 (do people call you ct?),

I started putting 2 and 2 and 3 together last night, and it's a "kick yourself" solution. I was pretty confident I had no significant changes on my Gentoo server, but I couldn't get large file transfers working from 2x Windows 10 clients, 1x Windows 7 client, and 1x OSX client.

I ran testparm, and smb.conf was as I remembered - very simple. Finally, I noticed that my ssh sessions had been kind of laggy recently, so I did some pinging. It seemed normal at <0.3 ms, then...

20% packet loss in one direction (client to server). In the other direction (server to client), even stranger. I could always get 12-16 good pings, then 100% loss for any more after that. I can only guess that a dozen or so packets was enough to shift 50-100 kB files over samba, but anything larger hit the wall.

So, now one of my Gigabit switches is sitting in the corner, thinking about what a bad boy it's been. I've been digging through forums, bug-trackers and config files for days!!! With this out of the way...I'm getting 75 MB/s after protocol etc.

With respect to the OP though, i can finally report "no change" with kernel 4.9 on my system. (As a side note...who makes good quality, basic switches?)
Back to top
View user's profile Send private message
kilburna
Tux's lil' helper
Tux's lil' helper


Joined: 20 Sep 2004
Posts: 107

PostPosted: Thu Mar 23, 2017 12:17 am    Post subject: Reply with quote

I have the exact problem after switching from kernel 4.4 to 4.9. 4.4 was working perfectly for months. Since switching to 4.9, I am getting timeouts and failures randomly. This is not hardware related as I have new equipment and the kernel change was the only thing done.

I read another article that cifs has changed a bit since since 4.4. I have not yet found a solution.

I have 4.9 cifs slower than 4.4.
Code:

//192.168.5.1/Sales on /mnt/sales type cifs (rw,relatime,vers=1.0,cache=strict,username=XXXX,domain=XXXXX,uid=0,noforceuid,gid=0,noforcegid,
addr=192.168.5.1,file_mode=0755,dir_mode=0755,nounix,serverino,mapposix,rsize=61440,wsize=16580,echo_interval=60,actimeo=1)


[Moderator edit: added [code] tags to preserve output layout; broke long whitespace-free line to fix thread layout. -Hu]
Back to top
View user's profile Send private message
ct85711
Veteran
Veteran


Joined: 27 Sep 2005
Posts: 1701

PostPosted: Thu Mar 23, 2017 2:22 am    Post subject: Reply with quote

Just out of curiosity, see if the problem exists on say either a 4.8 or a 4.10 kernel. I noticed a couple bug reports on Fedora indicating a possible issue with the power safe mode. Now, I don't know how much I trust those bug reports as I didn't see anything else pop up indicating the same thing. Either way, testing a different kernel should help us confirm or deny that the kernel is the cause.
Back to top
View user's profile Send private message
axl
l33t
l33t


Joined: 11 Oct 2002
Posts: 637
Location: Romania

PostPosted: Fri Mar 24, 2017 5:27 am    Post subject: Reply with quote

I would test if other protocols are faster than cifs. nfs scp ftp come to mind. those are very fast and easy to test. it's a good way to determine if the line is capable of higher speeds.

second, i'm not entirely sure if the problem originates from server or client. kernel 4.4/4.8 i assume is the clients kernel. the one doing the mount. but perhaps the server changed too. any detail on the server part?

and third, I had to stick to 4.4 because of an acpi bug that made the system limp with one core. 4.8/9/10 now series is the second most bleeding edge type kernel they have.

usually when linus releases a kernel from 4.4 series, he also releases a version of the 4.8/9/10 series. same bugfixes. some differences ofc. like today. 4.10.5. 4.4.56.
Back to top
View user's profile Send private message
Display posts from previous:   
Reply to topic    Gentoo Forums Forum Index Kernel & Hardware All times are GMT
Page 1 of 1

 
Jump to:  
You cannot post new topics in this forum
You cannot reply to topics in this forum
You cannot edit your posts in this forum
You cannot delete your posts in this forum
You cannot vote in polls in this forum