Re: More robust filesystem?
In article <[EMAIL PROTECTED]>, Wayne Cuddy <[EMAIL PROTECTED]> wrote: >On Wed, 14 Oct 1998 [EMAIL PROTECTED] wrote: >> Maybe the local drive is read-only and the writable stuff is on a server. > >They must have a haul-ass network since netscape uses the cache >extensively. With lots of people browsing this could kill bandwidth. On all machines here the netscape local cache is set to 0KB. All browsers use a central squid caching proxy which is _much_ faster than the netscape internal disk cache anyway Mike. -- "Did I ever tell you about the illusion of free will?" -- Sheriff Lucas Buck, ultimate BOFH.
Re: More robust filesystem?
On Wed, 14 Oct 1998 [EMAIL PROTECTED] wrote: > Maybe the local drive is read-only and the writable stuff is on a server. > > Paul They must have a haul-ass network since netscape uses the cache extensively. With lots of people browsing this could kill bandwidth. > > > > -- > Unsubscribe? mail -s unsubscribe [EMAIL PROTECTED] < /dev/null > Wayne Cuddy CRB-WEB (C & H Consulting) http://www.crb-web.com [EMAIL PROTECTED]
Re: More robust filesystem?
On Wed, 14 Oct 1998, Jim McCloskey wrote: > |> Then people will use the power off switch. And I can't cut that off > |> because it's used to switch the workstations off when the lab closes! > > I happened to be in the public library of the Beaubourg in Paris this > summer, looking for some free net access. > > They provide a roomfull of machines from which people can use > netscape, telnet and so on for free. I was interested to notice that > all of these machines were running Linux (don't know what > flavour). And I was fascinated and horrified to notice that every user > ended their session by simply hitting the power-off button. Finesse > gallique. The machines rebooted without complaint every time > though. Each machine must have this happen to it at least 30 or 40 > times a day. > > So there must be some way to do it, > > Jim Maybe the local drive is read-only and the writable stuff is on a server. Paul
Re: More robust filesystem?
|> Then people will use the power off switch. And I can't cut that off |> because it's used to switch the workstations off when the lab closes! I happened to be in the public library of the Beaubourg in Paris this summer, looking for some free net access. They provide a roomfull of machines from which people can use netscape, telnet and so on for free. I was interested to notice that all of these machines were running Linux (don't know what flavour). And I was fascinated and horrified to notice that every user ended their session by simply hitting the power-off button. Finesse gallique. The machines rebooted without complaint every time though. Each machine must have this happen to it at least 30 or 40 times a day. So there must be some way to do it, Jim
Re: More robust filesystem?
Stephen J. Carpenter writes: > 1) Stop it from happening... > cut the wires that goto the reset button...replace it with a key > switch other bits of wiring... Then people will use the power off switch. And I can't cut that off because it's used to switch the workstations off when the lab closes! > 2) mount as much as possible read-only. (/usr /etc ) This is probably what I will be doing. > 3) On a network? NFS mounts don't seem to mind this abuse at all > is NFS-root not an option? Not really - we have big enough HDs on each station, and it would be a shame to waste them. Besides, we don't want to buy a new server for the NFS-root. -- Alex Shnitman [EMAIL PROTECTED], [EMAIL PROTECTED] http://alexsh.home.ml.org
Re: More robust filesystem?
Alex Shnitman <[EMAIL PROTECTED]> writes: > Hi. > > We're using Debian workstations in our labs, and as expected they > rarely get shut down properly, many times they are just reset or > switched off, either due to ignorance or not caring. The question is > whether there is a way to configure the kernel to issue updates to the > meta-data more frequently, even in expense of performance? Or what > else can I do to keep the filesystems on the workstations more stable, > in addition to user educating on which we're of course working? Do a "man chattr". It lets you set specific bits in an ext2 filesystem. One of the attributes that you can set is the "S" flag, which tells the filesystem to do synchronous updates on this file or directory. It also mentions that "mount" has a "sync" option. Later, Dale -- + finger for pgp public key -+ | Dale E. Martin | Clifton Labs, Inc. | Senior Computer Engineer| | [EMAIL PROTECTED]|http://www.clifton-labs.com | +--+
Re: More robust filesystem?
In article <[EMAIL PROTECTED]>, Alex Shnitman <[EMAIL PROTECTED]> wrote: >We're using Debian workstations in our labs, and as expected they >rarely get shut down properly, many times they are just reset or >switched off, either due to ignorance or not caring. Wire the reset button through the keylock (so you can still use reset if needed), and disable the power button. With ATX you can even disable turning off the power through the powerbutton in the BIOS. If your kernel is compiled with the right options, shutdown -h (halt) will poweroff after shutdown. So put "-h" in the ctrlaltdel entry in /etc/inittab instead of -r so users can shutdown & turn off the machine with ctrl-alt-del Mike. -- "Did I ever tell you about the illusion of free will?" -- Sheriff Lucas Buck, ultimate BOFH.
Re: More robust filesystem?
On Wed, Oct 14, 1998 at 03:50:20PM +0200, Alex Shnitman wrote: > Peter Iannarelli writes: > > > You could put a directive in your crontab to issue a sync > > every 5 minutes of every hour of every day. > > That's not quite the issue - Linux syncronizes its buffers whenever it > has a chance anyway. What I'd like to know is whether there is a way > to minimize the damage in the case of a "reset" when the machine was > busy. hmm A few ideas come to mind... 1) Stop it from happening... cut the wires that goto the reset button...replace it with a key switch other bits of wiring... 2) mount as much as possible read-only. (/usr /etc ) 3) On a network? NFS mounts don't seem to mind this abuse at all is NFS-root not an option? not too much I can think of other than hacking your kernel... -Steve -- /* -- Stephen Carpenter <[EMAIL PROTECTED]> --- <[EMAIL PROTECTED]> */ E-mail "Bumper Stickers": "A FREE America or a Drug-Free America: You can't have both!" "honk if you Love Linux"
RE: More robust filesystem?
Is anybody working on a journalling/resilient file system for Linux? Rgds, Andy Chittenden Email: [EMAIL PROTECTED] Fax:+44 1753 661011 Multimedia Development, Madge Networks Ltd Wexham Springs, Framewood Road, Wexham, Slough SL3 6PJ, England
Re: More robust filesystem?
Peter Iannarelli writes: > You could put a directive in your crontab to issue a sync > every 5 minutes of every hour of every day. That's not quite the issue - Linux syncronizes its buffers whenever it has a chance anyway. What I'd like to know is whether there is a way to minimize the damage in the case of a "reset" when the machine was busy. -- Alex Shnitman [EMAIL PROTECTED], [EMAIL PROTECTED] http://alexsh.home.ml.org
Re: More robust filesystem?
Hello Alex: You could put a directive in your crontab to issue a sync every 5 minutes of every hour of every day. Peter Alex Shnitman wrote: > Hi. > > We're using Debian workstations in our labs, and as expected they > rarely get shut down properly, many times they are just reset or > switched off, either due to ignorance or not caring. The question is > whether there is a way to configure the kernel to issue updates to the > meta-data more frequently, even in expense of performance? Or what > else can I do to keep the filesystems on the workstations more stable, > in addition to user educating on which we're of course working? > > -- > Alex Shnitman > [EMAIL PROTECTED], [EMAIL PROTECTED] > http://alexsh.home.ml.org > > -- > Unsubscribe? mail -s unsubscribe [EMAIL PROTECTED] < /dev/null begin: vcard fn: Peter Iannarelli n: Iannarelli;Peter org:GenX Internet Laboratories Inc. adr:20 Madison Ave.;;;Toronto;Ontario;M5R 2S1;Canada email;internet: [EMAIL PROTECTED] title: Engineer tel;work: 1+ 416 929 1885 tel;fax:1+ 416 929 1056 note: Unix/Linux Support x-mozilla-cpt: ;0 x-mozilla-html: FALSE version:2.1 end:vcard
More robust filesystem?
Hi. We're using Debian workstations in our labs, and as expected they rarely get shut down properly, many times they are just reset or switched off, either due to ignorance or not caring. The question is whether there is a way to configure the kernel to issue updates to the meta-data more frequently, even in expense of performance? Or what else can I do to keep the filesystems on the workstations more stable, in addition to user educating on which we're of course working? -- Alex Shnitman [EMAIL PROTECTED], [EMAIL PROTECTED] http://alexsh.home.ml.org