On 12/16/20 9:12 AM, Satvinder Singh wrote:
You disabled on the bacula VM? I tried on th windows VMs but that didn’t help.

  <https://htmlsig.com/t/000001EVKWBG>


Yes. On the interface used by bacula-sd.




Satvinder Singh / Operations Manager
ssi...@celerium.com / Cell: 703-989-8030

Celerium
Office: 703-418-6315
www.celerium.com <http://www.celerium.com/>

  <https://htmlsig.com/t/000001EWFY3Y>   <https://htmlsig.com/t/000001ETVPF1>   
<https://htmlsig.com/t/000001EP3BJ4>


On 12/16/20, 9:09 AM, "Josh Fisher" <jfis...@jaybus.com> wrote:

     The message has originated from an External Source. Please use caution 
when opening attachments, clicking links, or responding to this email.

     The same happened to me. I posted here on August 5 about debugging it. I
     believe it to be a bug in virtio-net or in the bridge code that can in
     some cases cause failures when a physical NIC (or at least igb driver)
     with some combination of offloading enabled is attached to the same
     bridge as a virtio-net device. I found that disabling generic receive
     offload, TCP  segmentation offload, and generic segmentation offload on
     the physical NIC made the problem go away.

     I never found the time to experiment further to see if it was one
     offload feature in particular or some combination that caused the issue.
     In any case, the error is a rare occurrence, because I could never
     trigger the issue with iperf, or with an incremental backup. Only a full
     backup large enough to run for at least half an hour would fail. With
     those offloading features disabled on the SD's interface I haven't had a
     Bacula job failure in months.


     On 12/16/20 8:44 AM, Satvinder Singh wrote:
     > HI what I ended up doing was rolling back to a previous kernel version 
on Bacula VM and that resolved the issue, still trying to figure out what in the 
kernel update could have caused the issue.
     >
     > Thanks
     >
     >   <https://htmlsig.com/t/000001EVKWBG>
     >
     >
     > Satvinder Singh / Operations Manager
     > ssi...@celerium.com / Cell: 703-989-8030
     >
     > Celerium
     > Office: 703-418-6315
     > 
https://nam11.safelinks.protection.outlook.com/?url=http%3A%2F%2Fwww.celerium.com%2F&amp;data=04%7C01%7Cssingh%40celerium.com%7C72b60245b6ed4725a05608d8a1cc38a9%7Cae3f7f428b694faeaf693c125dee806d%7C0%7C0%7C637437245799143239%7CUnknown%7CTWFpbGZsb3d8eyJWIjoiMC4wLjAwMDAiLCJQIjoiV2luMzIiLCJBTiI6Ik1haWwiLCJXVCI6Mn0%3D%7C1000&amp;sdata=mS6Km9T1Ui5vYPRU0KvULQPZph2COMUmw0mJtUcXWB4%3D&amp;reserved=0
 
<https://nam11.safelinks.protection.outlook.com/?url=http%3A%2F%2Fwww.celerium.com%2F&amp;data=04%7C01%7Cssingh%40celerium.com%7C72b60245b6ed4725a05608d8a1cc38a9%7Cae3f7f428b694faeaf693c125dee806d%7C0%7C0%7C637437245799143239%7CUnknown%7CTWFpbGZsb3d8eyJWIjoiMC4wLjAwMDAiLCJQIjoiV2luMzIiLCJBTiI6Ik1haWwiLCJXVCI6Mn0%3D%7C1000&amp;sdata=mS6Km9T1Ui5vYPRU0KvULQPZph2COMUmw0mJtUcXWB4%3D&amp;reserved=0>
     >
     >   <https://htmlsig.com/t/000001EWFY3Y>   <https://htmlsig.com/t/000001ETVPF1>  
 <https://htmlsig.com/t/000001EP3BJ4>
     >
     >
     > On 12/16/20, 8:42 AM, "Josh Fisher" <jfis...@jaybus.com> wrote:
     >
     >      The message has originated from an External Source. Please use 
caution when opening attachments, clicking links, or responding to this email.
     >
     >      I'm late answering this, but I suggest trying again with some TCP
     >      offloading features disabled on the interface used by SD.
     >
     >      /sbin/ethtool -K ethX tso off gso off gro off
     >
     >      There are indeed buggy drivers that fail to properly support the
     >      (virtual?) hardware offload features, including at least some 
version(s)
     >      of virtio-net, as I discovered and wrote about in my August 5 2020 
post
     >      here. Keep in mind that Bacula, of necessity, taxes the network for 
long
     >      periods of time, often for hours. It reveals network issues that a 
few
     >      seconds or minutes of iperf cannot. (In fact, I think virtio 
developers
     >      should consider using bacula-fd in a VM in their 
destructive/performance
     >      testing.)
     >
     >
     >      On 12/8/20 2:55 PM, Satvinder Singh wrote:
     >      > Hi,
     >      >
     >      > I have been testing out bacula for past few weeks. I have setup 
jobs for linux and windows clients (server 2016 and server 2019). The linux backups 
are running well, with almost gigabit speeds as we have a gigabit backbone network. 
But then windows backups are extremely slow averaging around 65kbps. I have tested 
the network connectivity between the backup server and clients using iperf and there 
is no issue there I see speeds of over 700mbps. I have also tried disabling VSS and 
enabling Spool Data but no change. I have also tried the Maximum Network Buffer Size 
= 32768 on the client but still no change. There is no firewall running on the 
windows machine.
     >      >
     >      > Has anyone seen this any help is greatly appreciated.
     >      >
     >      > Thanks
     >      > Satvinder
     >      >
     >      > Disclaimer: This message is intended only for the use of the 
individual or entity to which it is addressed and may contain information which is 
privileged, confidential, proprietary, or exempt from disclosure under applicable 
law. If you are not the intended recipient or the person responsible for delivering 
the message to the intended recipient, you are strictly prohibited from disclosing, 
distributing, copying, or in any way using this message. If you have received this 
communication in error, please notify the sender and destroy and delete any copies 
you may have received.
     >      >
     >      > _______________________________________________
     >      > Bacula-users mailing list
     >      > Bacula-users@lists.sourceforge.net
     >      > https://lists.sourceforge.net/lists/listinfo/bacula-users
     >
     > Disclaimer: This message is intended only for the use of the individual 
or entity to which it is addressed and may contain information which is 
privileged, confidential, proprietary, or exempt from disclosure under applicable 
law. If you are not the intended recipient or the person responsible for 
delivering the message to the intended recipient, you are strictly prohibited from 
disclosing, distributing, copying, or in any way using this message. If you have 
received this communication in error, please notify the sender and destroy and 
delete any copies you may have received.

Disclaimer: This message is intended only for the use of the individual or 
entity to which it is addressed and may contain information which is 
privileged, confidential, proprietary, or exempt from disclosure under 
applicable law. If you are not the intended recipient or the person responsible 
for delivering the message to the intended recipient, you are strictly 
prohibited from disclosing, distributing, copying, or in any way using this 
message. If you have received this communication in error, please notify the 
sender and destroy and delete any copies you may have received.


_______________________________________________
Bacula-users mailing list
Bacula-users@lists.sourceforge.net
https://lists.sourceforge.net/lists/listinfo/bacula-users

Reply via email to