| Summary: | NFSv3 UPD low performance using dump | ||||||
|---|---|---|---|---|---|---|---|
| Product: | [Fedora] Fedora | Reporter: | Alessandro Selli <alessandroselli> | ||||
| Component: | nfs-utils | Assignee: | Steve Dickson <steved> | ||||
| Status: | CLOSED NOTABUG | QA Contact: | Fedora Extras Quality Assurance <extras-qa> | ||||
| Severity: | medium | Docs Contact: | |||||
| Priority: | unspecified | ||||||
| Version: | 15 | CC: | bfields, jlayton, steved | ||||
| Target Milestone: | --- | ||||||
| Target Release: | --- | ||||||
| Hardware: | i686 | ||||||
| OS: | Linux | ||||||
| Whiteboard: | |||||||
| Fixed In Version: | Doc Type: | Bug Fix | |||||
| Doc Text: | Story Points: | --- | |||||
| Clone Of: | Environment: | ||||||
| Last Closed: | 2011-07-06 10:46:48 UTC | Type: | --- | ||||
| Regression: | --- | Mount Type: | --- | ||||
| Documentation: | --- | CRM: | |||||
| Verified Versions: | Category: | --- | |||||
| oVirt Team: | --- | RHEL 7.3 requirements from Atomic Host: | |||||
| Cloudforms Team: | --- | Target Upstream Version: | |||||
| Attachments: |
|
||||||
|
Description
Alessandro Selli
2011-07-02 14:30:37 UTC
The writes are probably synchronous... Could you please post and trace of the over the wire traffic? Here is how: yum install wireshark tshark -w /tmp/data.pcap host <server> bzip2 /tmp/data.pcap Also it the client and server both Fedora 15? Created attachment 511433 [details]
wireshark-generated libcap file
Here is the data requested (bzip2-compressed file of the first 10MB of data). I confirm both the NFS client and the NFS server are Fedora15 boxes.
(In reply to comment #2) > Created attachment 511433 [details] > wireshark-generated libcap file > > Here is the data requested (bzip2-compressed file of the first 10MB of data). > I confirm both the NFS client and the NFS server are Fedora15 boxes. Well as expected the dump application is doing synchronous writes, note the FILE_SYNC flag set in all the WRITEs that are going over the wire. The last bullet in http://nfs.sourceforge.net/#faq_a1 has a good explanation FILE_SYNC writes. Do end the end there is not much can be done about the slowness, at least from an NFS point of view. I'm going to close this for now. If more information comes up please feel free to reopen it. |