Linux NFS development
 help / color / mirror / Atom feed
From: "Mkrtchyan, Tigran" <tigran.mkrtchyan@desy.de>
To: linux-nfs <linux-nfs@vger.kernel.org>
Cc: Mike Snitzer <snitzer@kernel.org>, trondmy <trondmy@kernel.org>,
	 anna <anna@kernel.org>
Subject: Re: [PATCH 0/2] pnfs/flexfiles: make DS connect application interrupt resistent
Date: Thu, 15 Oct 2026 11:58:08 +0200 (CEST)	[thread overview]
Message-ID: <1376472961.6381734.1792058288217.JavaMail.zimbra@desy.de> (raw)
In-Reply-To: <20261010221839.42172-1-tigran.mkrtchyan@desy.de>

[-- Attachment #1: Type: text/plain, Size: 1955 bytes --]


Dear NFS fellows,

I will appreciate comments on this series, including the separate krb5i handshake patch.

Thanks in advance,
   Tigran.

----- Original Message -----
> From: "Tigran Mkrtchyan" <tigran.mkrtchyan@desy.de>
> To: "trondmy" <trondmy@kernel.org>, "anna" <anna@kernel.org>
> Cc: "linux-nfs" <linux-nfs@vger.kernel.org>, "Mike Snitzer" <snitzer@kernel.org>, "Tigran Mkrtchyan"
> <tigran.mkrtchyan@desy.de>
> Sent: Sunday, 11 October, 2026 00:18:37
> Subject: [PATCH 0/2] pnfs/flexfiles: make DS connect application interrupt resistent

> In our testing we hit a situation where the pNFS flexfiles client
> goes into an infinite loop. The application spawns one reader process
> per CPU core to read multiple files in parallel, and kills the readers
> once a fixed time budget expires. The killed processes never exit.
> 
> We have identified three issues:
> 
> - Concurrent connections to data servers are serialized in
>  nfs4_discover_server_trunking() (not addressed yet).
> 
> - Regardless of the configured security flavor, the client always
>  tries krb5i first, addressed in a separate patch:
>  https://lore.kernel.org/linux-nfs/20261009155802.540179-1-tigran.mkrtchyan@desy.de/T/#u
> 
> - ff_layout_pg_init_read() and ff_layout_pg_init_write() retry in a
>  loop with ssleep() when no data server is available. Since ssleep()
>  is uninterruptible and the loop does not check for fatal signals,
>  a task with SIGKILL pending never leaves the loop.
> 
> These two patches address the last issue.
> 
> The patches tested on 7.3-rc3 and RHEL 5.14.0-757 kernels (with
> backport).
> 
> Tigranve  Mkrtchyan (2):
>  pNFS/flexfiles: make pg_init retry loops killable
>  pNFS/flexfiles: don't report an interrupted DS connect as a DS error
> 
> fs/nfs/flexfilelayout/flexfilelayout.c    | 23 ++++++++++++++++++++---
> fs/nfs/flexfilelayout/flexfilelayoutdev.c |  5 +++++
> 2 files changed, 25 insertions(+), 3 deletions(-)
> 
> --
> 2.56.0

[-- Attachment #2: S/MIME Cryptographic Signature --]
[-- Type: application/pkcs7-signature, Size: 2309 bytes --]

      parent reply	other threads:[~2026-10-15  9:58 UTC|newest]

Thread overview: 4+ messages / expand[flat|nested]  mbox.gz  Atom feed  top
2026-10-10 22:18 [PATCH 0/2] pnfs/flexfiles: make DS connect application interrupt resistent Tigran Mkrtchyan
2026-10-10 22:18 ` [PATCH 1/2] pNFS/flexfiles: make pg_init retry loops killable Tigran Mkrtchyan
2026-10-10 22:18 ` [PATCH 2/2] pNFS/flexfiles: don't report an interrupted DS connect as a DS error Tigran Mkrtchyan
2026-10-15  9:58 ` Mkrtchyan, Tigran [this message]

Reply instructions:

You may reply publicly to this message via plain-text email
using any one of the following methods:

* Save the following mbox file, import it into your mail client,
  and reply-to-all from there: mbox

  Avoid top-posting and favor interleaved quoting:
  https://en.wikipedia.org/wiki/Posting_style#Interleaved_style

* Reply using the --to, --cc, and --in-reply-to
  switches of git-send-email(1):

  git send-email \
    --in-reply-to=1376472961.6381734.1792058288217.JavaMail.zimbra@desy.de \
    --to=tigran.mkrtchyan@desy.de \
    --cc=anna@kernel.org \
    --cc=linux-nfs@vger.kernel.org \
    --cc=snitzer@kernel.org \
    --cc=trondmy@kernel.org \
    /path/to/YOUR_REPLY

  https://kernel.org/pub/software/scm/git/docs/git-send-email.html

* If your mail client supports setting the In-Reply-To header
  via mailto: links, try the mailto: link
Be sure your reply has a Subject: header at the top and a blank line before the message body.
This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox