[LU-8914] write call hung in wait_on_page_bit

• Status: Resolved
• Project: Lustre
• Type: Bug
• Priority: Minor
• Reporter: Olaf Faaland
• Assignee: Zhenyu Xu
• Resolution: Cannot Reproduce
• Labels: llnl

Environment

lustre-2.5.5-10chaos.4
rhel 7.3

• Severity: 3

Description

write system call hangs with identical stack to LU-4540.

Comments in that ticket ascribe the error to LU-3321 commits, however our patch stack does not have most of the LU-3321 commits. The only one I found is:

LU-4786 osc: to not pick busy pages for ELC

so it's unclear to me whether the problem we are seeing is due to the same root cause.

Comments

[Olaf Faaland]

See repository is named "lustre-release-fe-llnl" hosted on your gerritt server, for our exact code. This lustre was built from the head of branch 2.5.5-10chaos.

[Olaf Faaland]

gerrit link to the LU-3321 related change in our stack is
https://review.whamcloud.com/#/c/10795/

[Peter Jones]

Bobijam

Could you please assist with this one?

Thanks

Peter

[Zhenyu Xu]

Hi Jinshan,

I think the scenario is similar, ll_write_begin() grabbed a page lock and waiting for another page's WriteBack bit to be cleared, am I right?