Gentoo Forums
Gentoo Forums
Gentoo Forums
Quick Search: in
packet drops under vmware with newer kernels
View unanswered posts
View posts from last 24 hours

 
Reply to topic    Gentoo Forums Forum Index Networking & Security
View previous topic :: View next topic  
Author Message
tazinblack
Veteran
Veteran


Joined: 23 Jan 2005
Posts: 1146
Location: Baden / Germany

PostPosted: Wed Aug 14, 2013 10:20 am    Post subject: packet drops under vmware with newer kernels Reply with quote

Hi together,

I have an issue with some virtual gentoo machines running on vmware esxi 5.1 using newer kernels.
I run i.e icinga on these vms which sends and receives a lot of small network packets as I think.
Problem is that aI have a lot of dropped incoming packets:

Code:
enp2s0: flags=4163<UP,BROADCAST,RUNNING,MULTICAST>  mtu 1500
        inet <some ip>  netmask 255.255.255.0  broadcast <some bc adress>
        inet6 <some mac address>  prefixlen 64  scopeid 0x20<link>
        ether <some mac address>  txqueuelen 1000  (Ethernet)
        RX packets 16357  bytes 1872630 (1.7 MiB)
        RX errors 0  dropped 545  overruns 0  frame 0
        TX packets 17015  bytes 7781044 (7.4 MiB)
        TX errors 0  dropped 0 overruns 0  carrier 0  collisions 0


As you can see I have 545 dropped RX packets and this VM is only running for about 5 minutes.
This number increases every minute. After some days I have millions of them.
I also have very poor check latency in icinga. It does not matter if I use e1000 driver based network
interface or vmxnet3. It's the same behavior on both of them.

Looks like this is with kernels greater 3.2 On machines using old 2.6 kernels I do not have this issue.
I also have a ubuntu vm using 3.2.0.29 kernel which also drops packages but there I have only about 5600 over 239 days uptime and that I think may be normal.
Another suse sles11 with kernel 2.6.27.19-5-default does not drop any packet.

I tried to extent RX buffer with ethtool -G enp2s0 rx 4096 so that I have these values configured:

Code:
ethtool -g enp2s0
Ring parameters for enp2s0:
Pre-set maximums:
RX:             4096
RX Mini:        0
RX Jumbo:       0
TX:             4096
Current hardware settings:
RX:             4096
RX Mini:        0
RX Jumbo:       0
TX:             256


But this does not change anything. I also can not see dropped packets on the virtual or any physical switch.
Would be great if anyone out there could give me a hint how to fix this. :wink:
_________________
Gruß / Regards
tazinblack
_______________________________________________________
what's the point in being grown up if you can't be childish sometimes
Back to top
View user's profile Send private message
Neon Dude
n00b
n00b


Joined: 13 Jun 2004
Posts: 15
Location: London

PostPosted: Thu Aug 15, 2013 8:10 pm    Post subject: Reply with quote

Hi,

So dropped packets can take quite some debugging, but you may want to try experimenting with the net.core.netdev_max_backlog sysctl setting - try bumping it up and see if this has any impact. It may also require net.core.rmem_max/net.core.wmem_max to be increased too.

Good luck!
Back to top
View user's profile Send private message
tazinblack
Veteran
Veteran


Joined: 23 Jan 2005
Posts: 1146
Location: Baden / Germany

PostPosted: Fri Aug 16, 2013 6:18 am    Post subject: Reply with quote

Good idea, thank you!

The defaults were
Code:
sysctl net.core
net.core.dev_weight = 64
net.core.message_burst = 10
net.core.message_cost = 5
net.core.netdev_budget = 300
net.core.netdev_max_backlog = 1000
net.core.netdev_tstamp_prequeue = 1
net.core.optmem_max = 20480
net.core.rmem_default = 212992
net.core.rmem_max = 212992
net.core.rps_sock_flow_entries = 0
net.core.somaxconn = 128
net.core.warnings = 1
net.core.wmem_default = 212992
net.core.wmem_max = 212992
net.core.xfrm_acq_expires = 30
net.core.xfrm_aevent_etime = 10
net.core.xfrm_aevent_rseqth = 2
net.core.xfrm_larval_drop = 1


And I changed them to

Code:
sysctl net.core
net.core.dev_weight = 64
net.core.message_burst = 10
net.core.message_cost = 5
net.core.netdev_budget = 300
net.core.netdev_max_backlog = 10000
net.core.netdev_tstamp_prequeue = 1
net.core.optmem_max = 20480
net.core.rmem_default = 212992
net.core.rmem_max = 425984
net.core.rps_sock_flow_entries = 0
net.core.somaxconn = 128
net.core.warnings = 1
net.core.wmem_default = 212992
net.core.wmem_max = 425984
net.core.xfrm_acq_expires = 30
net.core.xfrm_aevent_etime = 10
net.core.xfrm_aevent_rseqth = 2
net.core.xfrm_larval_drop = 1


I also tried to adjust settings recommended here http://gnet.soe.ucsc.edu/doku.php?id=tcptuning and here http://www.speedguide.net/articles/linux-tweaking-121.

Unfortunately this does not have any effect.

I have round about 1 dropped packed per second and nothing I try seems to help :-(
_________________
Gruß / Regards
tazinblack
_______________________________________________________
what's the point in being grown up if you can't be childish sometimes
Back to top
View user's profile Send private message
tazinblack
Veteran
Veteran


Joined: 23 Jan 2005
Posts: 1146
Location: Baden / Germany

PostPosted: Thu Sep 19, 2013 10:59 am    Post subject: Reply with quote

isn't there anything new that can help?
_________________
Gruß / Regards
tazinblack
_______________________________________________________
what's the point in being grown up if you can't be childish sometimes
Back to top
View user's profile Send private message
Display posts from previous:   
Reply to topic    Gentoo Forums Forum Index Networking & Security 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