all lists on lists.proxmox.com
 help / color / mirror / Atom feed
From: "Fabian Grünbichler" <f.gruenbichler@proxmox.com>
To: Proxmox Backup Server development discussion
	<pbs-devel@lists.proxmox.com>,
	Stefan Reiter <s.reiter@proxmox.com>,
	Thomas Lamprecht <t.lamprecht@proxmox.com>
Subject: Re: [pbs-devel] [PATCH proxmox-backup 5/9] file-restore-daemon: work around tokio DuplexStream bug
Date: Fri, 07 May 2021 09:26:51 +0200	[thread overview]
Message-ID: <1620372217.h78x1o48e9.astroid@nora.none> (raw)
In-Reply-To: <07c15982-0adb-d3b6-02c3-560969df1779@proxmox.com>

On May 6, 2021 8:12 pm, Thomas Lamprecht wrote:
> On 06.05.21 17:26, Stefan Reiter wrote:
>> See this PR for more info: https://github.com/tokio-rs/tokio/pull/3756
>> 
>> As a workaround use a pair of connected unix sockets - this obviously
>> incurs some overhead, albeit not measureable on my machine. Once tokio
>> includes the fix we can go back to a DuplexStream for performance and
>> simplicity.
>> 
> 
> now that it got merged, do we want to cherry-pick this or await inclusion
> in stable instead of applying this workaround?
> 
> CC'ing Fabian as our tokio (or better said rust) packaging wizard

we can

A) use this workaround, and switch back once a fixed tokio version is 
available

B) cherry-pick the fix into our tokio package, not use this workaround, 
package and bump dependency ASAP once a tokio version with the fix is 
available

A has the advantage of not (subtly) breaking third-party builds that use 
crates from crates.io, but requires us to remember switching back once 
the fix is available. B has the advantage that we don't have to 
remember anything, since we'll update tokio sooner or later anyway ;)

both is fine for me, cherry-picking is about 10 minutes worth of work 
so..

> 
>> Signed-off-by: Stefan Reiter <s.reiter@proxmox.com>
>> ---
>> 
>> Thanks to @Wolfgang for helping debug this!
>> 
>>  src/bin/proxmox_restore_daemon/api.rs | 6 +++++-
>>  1 file changed, 5 insertions(+), 1 deletion(-)
>> 
>> diff --git a/src/bin/proxmox_restore_daemon/api.rs b/src/bin/proxmox_restore_daemon/api.rs
>> index f1d601ce..5aeb69f3 100644
>> --- a/src/bin/proxmox_restore_daemon/api.rs
>> +++ b/src/bin/proxmox_restore_daemon/api.rs
>> @@ -275,7 +275,11 @@ fn extract(
>>              bail!("file or directory {:?} does not exist", path);
>>          }
>>  
>> -        let (mut writer, reader) = tokio::io::duplex(1024 * 64);
>> +        // FIXME: DuplexStream is currently broken and doesn't wake pending writers on close, i.e.
>> +        // this doesn't drop the WatchdogInhibitor if we encounter an error (client aborts, etc...)
>> +        // see: https://github.com/tokio-rs/tokio/pull/3756
>> +        // let (mut writer, reader) = tokio::io::duplex(1024 * 64);
>> +        let (mut writer, reader) = tokio::net::UnixStream::pair()?;
>>  
>>          if pxar {
>>              tokio::spawn(async move {
>> 
> 
> 




  reply	other threads:[~2021-05-07  7:27 UTC|newest]

Thread overview: 17+ messages / expand[flat|nested]  mbox.gz  Atom feed  top
2021-05-06 15:26 [pbs-devel] [PATCH 0/9] Debug mode and smaller fixes for single file restore Stefan Reiter
2021-05-06 15:26 ` [pbs-devel] [PATCH proxmox-backup 1/9] file-restore: add debug mode with serial access Stefan Reiter
2021-05-06 15:26 ` [pbs-devel] [PATCH proxmox-backup 2/9] file-restore: try to kill VM when stale Stefan Reiter
2021-05-07  7:04   ` [pbs-devel] applied: " Thomas Lamprecht
2021-05-06 15:26 ` [pbs-devel] [PATCH proxmox-backup 3/9] file-restore: add more RAM for VMs with many drives or debug Stefan Reiter
2021-05-07  7:04   ` [pbs-devel] applied: " Thomas Lamprecht
2021-05-06 15:26 ` [pbs-devel] [PATCH proxmox-backup 4/9] file-restore: support more drives Stefan Reiter
2021-05-07  7:04   ` [pbs-devel] applied: " Thomas Lamprecht
2021-05-06 15:26 ` [pbs-devel] [PATCH proxmox-backup 5/9] file-restore-daemon: work around tokio DuplexStream bug Stefan Reiter
2021-05-06 18:12   ` Thomas Lamprecht
2021-05-07  7:26     ` Fabian Grünbichler [this message]
2021-05-06 15:26 ` [pbs-devel] [PATCH proxmox-backup 6/9] file-restore-daemon: watchdog: add inhibit for long downloads Stefan Reiter
2021-05-06 15:26 ` [pbs-devel] [PATCH proxmox-backup 7/9] file-restore-daemon: limit concurrent download calls Stefan Reiter
2021-05-06 15:26 ` [pbs-devel] [PATCH proxmox-backup-restore-image 8/9] kernel: power off on panic Stefan Reiter
2021-05-06 15:26 ` [pbs-devel] [PATCH proxmox-backup-restore-image 9/9] add debug initramfs as seperate package Stefan Reiter
2021-05-06 18:07   ` Thomas Lamprecht
2021-05-25 11:50 ` [pbs-devel] applied-series: [PATCH 0/9] Debug mode and smaller fixes for single file restore Thomas Lamprecht

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=1620372217.h78x1o48e9.astroid@nora.none \
    --to=f.gruenbichler@proxmox.com \
    --cc=pbs-devel@lists.proxmox.com \
    --cc=s.reiter@proxmox.com \
    --cc=t.lamprecht@proxmox.com \
    /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 an external index of several public inboxes,
see mirroring instructions on how to clone and mirror
all data and code used by this external index.
Service provided by Proxmox Server Solutions GmbH | Privacy | Legal