From: Robert Obkircher <r.obkircher@proxmox.com>
To: Christian Ebner <c.ebner@proxmox.com>, pbs-devel@lists.proxmox.com
Subject: Re: [PATCH v1 proxmox-backup 2/3] client: pxar: consistently ignore vanished files and warn about them
Date: Fri, 8 May 2026 13:40:05 +0200 [thread overview]
Message-ID: <89b54e4c-1f39-43d5-b3c2-6d03a6170c3d@proxmox.com> (raw)
In-Reply-To: <4d21540f-f92b-48e1-a586-81a2d4412b05@proxmox.com>
On 08.05.26 11:36, Christian Ebner wrote:
> On 5/7/26 5:48 PM, Robert Obkircher wrote:
>>
>> On 07.05.26 10:36, Christian Ebner wrote:
>>> Two nits inline, otherwise LGTM.
>>>
>>> Reviewed-by: Christian Ebner <c.ebner@proxmox.com>
>>>
>>> On 5/6/26 1:48 PM, Robert Obkircher wrote:
>>>> Combine the error paths and treat deleted entries equally in both
>>>> cases. Since this used to be a hard error in some cases, it also
>>>> makes
>>>> sense to emit a warning about it.
>>>>
>>>> Signed-off-by: Robert Obkircher <r.obkircher@proxmox.com>
>>>> ---
>>>> pbs-client/src/pxar/create.rs | 32
>>>> ++++++++++++--------------------
>>>> 1 file changed, 12 insertions(+), 20 deletions(-)
>>>>
>>>> diff --git a/pbs-client/src/pxar/create.rs
>>>> b/pbs-client/src/pxar/create.rs
>>>> index 08c5c9b44..47cbf50a8 100644
>>>> --- a/pbs-client/src/pxar/create.rs
>>>> +++ b/pbs-client/src/pxar/create.rs
>>>> @@ -695,36 +695,28 @@ impl Archiver {
>>>> })
>>>> });
>>>> - match match_result {
>>>> + let stat_result = match match_result {
>>>> Ok(Some(MatchType::Exclude)) => {
>>>> debug!("matched by exclude pattern
>>>> '{full_path:?}'");
>>>> continue;
>>>> }
>>>> - Ok(_) => (),
>>>> - Err(err) if err.not_found() => continue,
>>>> + Ok(_) => stat_result.map_or_else(do_stat, Ok),
>>>> + Err(e) => Err(e),
>>>> + };
>>>> +
>>>> + let stat = match stat_result {
>>>> + Ok(stat) => stat,
>>>> + Err(err) if err.not_found() => {
>>>> + warn!("warning: file vanished while reading
>>>> directory: {full_path:?}");
>>>
>>> nit: should use and slightly adapt the report_vanished_file() helper.
>> Imo the messages should be different, so this would essentially
>> turn into:
>>
>> fn report_vanished_file(message: &str, path: &Path) {
>> warn!("warning: file vanished {message}: {path:?}")
>> }
>>
>> which seemed overly complicated for something that would only be
>> called from two locations.
>
> Fine by me, but in that case I would suggest to cleanup and inline
> the other call site as well. No need to keep it around then ...
>
>>
>> I also noticed that `read_pxar_excludes` calls `open_file` without
>> setting `self.path`, so the existing helpers would print the wrong
>> message in that case.
>>
>>>
>>>> + continue;
>>>> + }
>>>> Err(Errno::ESTALE) => {
>>>>
>>>> self.report_stale_file_handle(Some(&full_path));
>>>> continue;
>>>> }
>>>> Err(err) => {
>>>> - return Err(err).with_context(|| format!("stat
>>>> failed on {full_path:?}"))
>>>> + return
>>>> Err(Error::from(err).context(format!("stat failed on
>>>> {full_path:?}")))
>>>
>>> nit: please use with_context() over context() here, since the former
>>> is evaluated lazily only once an error does occur [0].
>>>
>>> [0]
>>> https://docs.rs/anyhow/latest/anyhow/trait.Context.html#required-methods
>>>
>> How? This is the error path that converts the Errno to an anyhow
>> error. It doesn't call the method from that trait.
>
> diff --git a/pbs-client/src/pxar/create.rs
> b/pbs-client/src/pxar/create.rs
> index c101415c1..e1e4eea3b 100644
> --- a/pbs-client/src/pxar/create.rs
> +++ b/pbs-client/src/pxar/create.rs
> @@ -720,7 +720,8 @@ impl Archiver {
> continue;
> }
> Err(err) => {
> - return
> Err(Error::from(err).context(format!("stat failed on {full_path:?}")))
> + return Err(Error::from(err))
> + .with_context(|| format!("stat failed on
> {full_path:?}"))
> }
> };
>
> This should do, unless I'm overlooking something?
If we inline <Result as Context>::with_context then your version it
results in
match Err(Error::from(err)) {
Ok(ok) => Ok(ok),
Err(error) => Err(error.ext_context(context())),
}
which is not lazy. The original code had both versions and I chose the
more explicit one.
>
>>>
>>>> }
>>>> - }
>>>> -
>>>> - let stat = match stat_result {
>>>> - Some(stat) => stat,
>>>> - None => match do_stat() {
>>>> - Ok(stat) => stat,
>>>> - Err(Errno::ESTALE) => {
>>>> -
>>>> self.report_stale_file_handle(Some(&full_path));
>>>> - continue;
>>>> - }
>>>> - Err(err) => {
>>>> - return Err(
>>>> - Error::from(err).context(format!("stat
>>>> failed on {full_path:?}"))
>>>> - )
>>>> - }
>>>> - },
>>>> };
>>>> self.entry_counter += 1;
>>>
>
next prev parent reply other threads:[~2026-05-08 11:40 UTC|newest]
Thread overview: 13+ messages / expand[flat|nested] mbox.gz Atom feed top
2026-05-06 11:49 [PATCH v1 proxmox-backup 0/3] improve fstatat error handling Robert Obkircher
2026-05-06 11:49 ` [PATCH v1 proxmox-backup 1/3] client: pxar: improve variable names Robert Obkircher
2026-05-07 8:38 ` Christian Ebner
2026-05-06 11:49 ` [PATCH v1 proxmox-backup 2/3] client: pxar: consistently ignore vanished files and warn about them Robert Obkircher
2026-05-07 8:38 ` Christian Ebner
2026-05-07 15:50 ` Robert Obkircher
2026-05-08 9:38 ` Christian Ebner
2026-05-08 11:40 ` Robert Obkircher [this message]
2026-05-08 13:26 ` Christian Ebner
2026-05-08 14:03 ` Robert Obkircher
2026-05-08 14:12 ` Christian Ebner
2026-05-06 11:49 ` [PATCH v1 proxmox-backup 3/3] client: pxar: skip files on fstatat permission errors Robert Obkircher
2026-05-07 8:42 ` Christian Ebner
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=89b54e4c-1f39-43d5-b3c2-6d03a6170c3d@proxmox.com \
--to=r.obkircher@proxmox.com \
--cc=c.ebner@proxmox.com \
--cc=pbs-devel@lists.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 a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox