linux-hotplug.vger.kernel.org archive mirror
 help / color / mirror / Atom feed
From: "Kevin P. Fleming" <kpfleming@backtobasicsmgmt.com>
To: linux-hotplug@vger.kernel.org
Subject: Re: udev hangs under high loads
Date: Sun, 24 Oct 2004 19:27:30 +0000	[thread overview]
Message-ID: <417C0222.3080402@backtobasicsmgmt.com> (raw)
In-Reply-To: <20041023054119.GA11915@kroah.com>

Olaf Hering wrote:

> I had not time yet to look at the udev locking bugs.

I too would be unhappy if udev was wasting this much memory for extra 
files related to my /dev nodes (I moved to udev precisely to avoid 
problems like this <G>).

Not having looked at the code my suggestion may be way off-base, but 
wouldn't a very simple database that supports _only_ append operations 
(no in-place editing) suffice for udev's needs? There are many 
applications (BIND, Zope, etc.) that work just fine with databases like 
this, where any new or changed record is appended to the file, then 
pointers are updated to reflect the new record's presence. Seems to me 
that given the very small amount of update activity to udev's database 
(for most systems it will be 95+% appends when the system boots, and 
very little activity after that) that the small amount of wasted space 
due to "dead" records being in the database would be acceptable.

Then again, some of the threads about this issue seemed to relate to 
multiple systems sharing the same /dev directory over NFS, which I don't 
understand at all... Why would someone even want to do that (i.e. what 
would the value be)? If changes were made to handle non-locking-capable 
mounts for /dev, IMHO those changes should be reverted, and the 
documentation updated to say that if someone wants to use udev in such a 
situation, they are on their own :-)


-------------------------------------------------------
This SF.net email is sponsored by: IT Product Guide on ITManagersJournal
Use IT products in your business? Tell us what you think of them. Give us
Your Opinions, Get Free ThinkGeek Gift Certificates! Click to find out more
http://productguide.itmanagersjournal.com/guidepromo.tmpl
_______________________________________________
Linux-hotplug-devel mailing list  http://linux-hotplug.sourceforge.net
Linux-hotplug-devel@lists.sourceforge.net
https://lists.sourceforge.net/lists/listinfo/linux-hotplug-devel

  parent reply	other threads:[~2004-10-24 19:27 UTC|newest]

Thread overview: 17+ messages / expand[flat|nested]  mbox.gz  Atom feed  top
2004-10-23  5:41 udev hangs under high loads Greg KH
2004-10-23  6:31 ` Kay Sievers
2004-10-24  5:08 ` Kay Sievers
2004-10-24  5:53 ` Olaf Hering
2004-10-24 15:53 ` Kay Sievers
2004-10-24 15:57 ` Olaf Hering
2004-10-24 19:27 ` Kevin P. Fleming [this message]
2004-10-24 19:46 ` Kay Sievers
2004-10-24 19:51 ` Marco d'Itri
2004-10-24 19:56 ` Kay Sievers
2004-10-25  2:51 ` Kay Sievers
2004-10-25  7:01 ` Greg KH
2004-10-25 17:07 ` Kay Sievers
2004-10-25 17:12 ` Kevin P. Fleming
2004-10-25 17:27 ` Nishanth Aravamudan
2004-10-25 17:53 ` Kay Sievers
2004-10-25 20:55 ` Nishanth Aravamudan

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=417C0222.3080402@backtobasicsmgmt.com \
    --to=kpfleming@backtobasicsmgmt.com \
    --cc=linux-hotplug@vger.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;
as well as URLs for NNTP newsgroup(s).