public inbox for elfutils@sourceware.org
 help / color / mirror / Atom feed
From: "fche at redhat dot com" <sourceware-bugzilla@sourceware.org>
To: elfutils-devel@sourceware.org
Subject: [Bug debuginfod/27918] New: add control file for triggered traversal
Date: Wed, 26 May 2021 20:13:55 +0000	[thread overview]
Message-ID: <bug-27918-10460@http.sourceware.org/bugzilla/> (raw)

https://sourceware.org/bugzilla/show_bug.cgi?id=27918

            Bug ID: 27918
           Summary: add control file for triggered traversal
           Product: elfutils
           Version: unspecified
            Status: NEW
          Severity: normal
          Priority: P2
         Component: debuginfod
          Assignee: unassigned at sourceware dot org
          Reporter: fche at redhat dot com
                CC: elfutils-devel at sourceware dot org
  Target Milestone: ---

On a large debuginfod installation, where a full traversal might take hours but
things hardly ever change, it would help to focus debuginfod's traversal to
directories known to have changed.  For example, an external monitor of a build
system could notify us that a new build XYZ exists, and map that to a likely
subdirectory.

More specifically, let debuginfod be run with a --traversal-trigger=FILENAME
option.  In the traversal thread that normally sleeps quite a while, let it
check for the existence & nonemptiness of this file.  If it exists & has
content, the traversal thread should immediately begin a traversal/scan process
for the named directories, as though they were given on command line, probably
into some sibling vector to source_paths[].  Then it should truncate the file
(if is still unmodified from the beginning of the read operation - so as to
avoid losing data on a race).

If a traversal is already in progress, so the traversal thread is sitting
inside scan_source_paths()'s fts loop, there probably isn't a super simple way
of suspending that search, focusing on the new directories, then resuming ...
so probably not worth accelerating this case.  We could let the fts finish, and
then let the thread process the trigger file back out there.

An external program could  "echo /path/ >> FILENAME"  whenever, and debuginfod
should have that content scanned rapidly.  With that in place, such sites might
choose a very long -tNNN time for thorough full traversals.

-- 
You are receiving this mail because:
You are on the CC list for the bug.

             reply	other threads:[~2021-05-26 20:13 UTC|newest]

Thread overview: 5+ messages / expand[flat|nested]  mbox.gz  Atom feed  top
2021-05-26 20:13 fche at redhat dot com [this message]
2021-06-08 14:12 ` [Bug debuginfod/27918] " dichen at redhat dot com
2021-06-08 14:13 ` dichen at redhat dot com
2022-07-14 16:30 ` nsanci at redhat dot com
2022-07-14 16:30 ` nsanci at redhat dot com

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=bug-27918-10460@http.sourceware.org/bugzilla/ \
    --to=sourceware-bugzilla@sourceware.org \
    --cc=elfutils-devel@sourceware.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;
as well as URLs for read-only IMAP folder(s) and NNTP newsgroup(s).