Dear Managers,
Several days ago I posted the following queries and received
the indicated responses:
> 1) I use the "trashcan" utilities in AdvFS to provide undelete
>features for my users. This morning one of my users indicated his
>trashcan was no longer working. I checked all users: approximately
>2/3 of the accounts no longer had attached trashcans, but the other
>1/3 were OK (including this account, which is why I didn't notice it
>myself!) I reattached the trashcans, but can anyone think of a reason
>why this "unattachment" took place?
No responses were received on this one. I have installed a
cron job that runs once a month to reattach unattached trashcans.
This is treating the symptoms instead of the disease, but I don't know
what else to do.
> 2) A recent post mentioned using the -N switch to vdump to
>store multiple savesets on a tape (or using /dev/nrmt0h). How do you
>recover them with vrestore? I could only find the first one.
Responses were received from:
David Gadbois <gadbois_at_cyc.com>
Ron Barrett <ron_barrett_at_corp.cubic.com>
Dave Wolinski <wolinski_at_umalp1.physics.lsa.umich.edu>
Alan Rollow <alan_at_nabeth.cxo.dec.com>
Winfried Huber <win_at_tukan.ffb.eunet.de>
All respondents suggested using mt fsf [count]. Sorry, I was
so focussed on vrestore I missed this simple command!
> 3) Several posts last year mentioned problems with sendmail
>after applying the syslogd patch (ECO SSRT0359_OSF1032C). I would
>like to perform the patch, but not if it will disrupt my E-mail
>service. Any further word from DEC or anyone else?
Responses were received from:
Arrigo Triulzi <arrigo_at_lpac.ac.uk>
Dave Wolinski <wolinski_at_umalp1.physics.lsa.umich.edu>
In short: DON'T PATCH!
> 4) We use disk quotas on /var to prevent huge amounts of
>unread mail from piling up (a relatively low soft limit and a high
>hard limit to accommodate transient use). Since this is an educational
>institution we also have a high rate of user turnover. Is there any
>way of DELETING a user from the quota database?
Only Alan Rollow (above) responded on this one. I'll let him
speak for himself:
The software that creates the quota files, creates one entry
for every potential ID in the password and group file. It
uses the highest UID (or GID) to determine how many entries
to create. If there is a large gap between assigned IDs then
the software creates a sparse file with data only allocated to
the parts of the file that need it. Unless you're deleting
IDs off the end, an entry (space allocated or otherwise) will
always be present for an ID. If you don't want the quotas for
an ID enforced, change the limits to zero.
It is worth noting that in password and group files that have
the ID for "nobody" set to -2, this usually sets the upper bound
on the IDs. It gets converted to an unsigned 16 or 32 bit number
and the number of entries are set accordingly.
If you delete and don't reuse large blocks of IDs, then you need to
find a way to get the file reallocated. For Advfs, I don't think
you can do this short of backing up, recreating the domain and
file sets and restoring the data. For UFS, I it is simply a
matter of turning off quotas, deleting the old quota file, creating
a new empty one and rebuilding the quotas. This has the problem
of not preserving the quotas for the remaining users. Since it
is possible to write a new quota file easily for UFS, you could
also use dd(1) or a custom copying program to only copy the parts
of the file that are interesting. The dd(1) works when you've trimmed
a lot of IDs off the end and can shrink the file.
(Larry again) Thanks for all the help.
Larry Griffith
============================================================================
Larry Griffith Dept. of Computer & Info Science
larry_at_garfield.wsc.mass.edu Westfield State College
(413) 572-5294 Westfield, MA 01086 USA
PGP public key available at:
http://garfield.wsc.mass.edu/dcis/griffith.html
============================================================================
Received on Tue Jan 16 1996 - 14:59:35 NZDT