Re: Staying -current with cvsup or cvsync
On 2014-12-02, Jungle Boogie jungleboog...@gmail.com wrote: Dear Stuart, From: Stuart Henderson s...@spacehopper.org Sent: Tue, 2 Dec 2014 10:40:22 + (UTC) To: misc@openbsd.org Subject: Re: Staying -current with cvsup or cvsync On 2014-11-28, Jungle Boogie jungleboog...@gmail.com wrote: Hello All, For the last several updates I've applied to my system, I've used plain CVS: cvs -q up -Pd This is pretty slow for some reason, but I understand that's just how CVS works. I just timed an update of /usr/ports on my laptop at 63 seconds. That's fetching from a good anoncvs server, with /usr/ports on SSD and mounted like this /dev/sd1j on /usr/ports type ffs (local, noatime, nodev, nosuid, softdep) 63 seconds is quite impressive! I've got a pata drive with only: (local, nodev) softdep can help a lot with big cvs updates, especially on disks which are slower to access. Lots of files involved in a ports or src cvs tree (especially ports) so there are a large number of inode changes that need to be written to disk, How often do you fetch/rebuild? It varies, I probably update the entire ports tree on my laptop once or twice a week, and smaller parts if I'm working on them or if I see an update I want in the commit log. For base, the last full update I did was about 10 days ago, but again I've updated smaller parts more often and I often update the kernel every few days. There will be lots of differences between people (and at different times depending on what they're working on).
Re: Staying -current with cvsup or cvsync
On 2014-11-28, Jungle Boogie jungleboog...@gmail.com wrote: Hello All, For the last several updates I've applied to my system, I've used plain CVS: cvs -q up -Pd This is pretty slow for some reason, but I understand that's just how CVS works. I just timed an update of /usr/ports on my laptop at 63 seconds. That's fetching from a good anoncvs server, with /usr/ports on SSD and mounted like this /dev/sd1j on /usr/ports type ffs (local, noatime, nodev, nosuid, softdep) Does this mean cvssup is no longer used? Correct, the server side was written in Modula-3 which on OpenBSD has only ever been ported to i386 (most anoncvs servers are now running amd64) and it was not widely used, so wasn't worth the maintenance headache and extra exposure on servers. Then I came across cvsync: http://www.openbsd.org/cvsync.html Is cvsync preferred now? CVSup was able to either mirror the full repository, or make a checkout. The method you were looking at for CVSup was just for making a checkout. (This was quite widely used by FreeBSD in the past, but in OpenBSD the main method of users fetching the tree was from anoncvs mirrors). cvsync is only used for mirroring the full repo. Useful if connectivity between you and an anoncvs mirror isn't very fast, or if you want to hack offline and still be able to make diffs etc. Unlike CVSup it cannot do a direct checkout. I used to run a local cvsync mirror at home. But then the anoncvs server I used had some upgrades and got much faster so I now just fetch directly from there, unless I am going to be travelling and want an up-to-date local copy of the tree on my laptop.
Re: Staying -current with cvsup or cvsync
Dear Stuart, From: Stuart Henderson s...@spacehopper.org Sent: Tue, 2 Dec 2014 10:40:22 + (UTC) To: misc@openbsd.org Subject: Re: Staying -current with cvsup or cvsync On 2014-11-28, Jungle Boogie jungleboog...@gmail.com wrote: Hello All, For the last several updates I've applied to my system, I've used plain CVS: cvs -q up -Pd This is pretty slow for some reason, but I understand that's just how CVS works. I just timed an update of /usr/ports on my laptop at 63 seconds. That's fetching from a good anoncvs server, with /usr/ports on SSD and mounted like this /dev/sd1j on /usr/ports type ffs (local, noatime, nodev, nosuid, softdep) 63 seconds is quite impressive! I've got a pata drive with only: (local, nodev) How often do you fetch/rebuild? I plan on making a low power router (not really looking at the APU devices anymore) and in that, I'll use SSD or msata. Does this mean cvssup is no longer used? Correct, the server side was written in Modula-3 which on OpenBSD has only ever been ported to i386 (most anoncvs servers are now running amd64) and it was not widely used, so wasn't worth the maintenance headache and extra exposure on servers. Well the book I referenced is from 2003, when i386 was common and 3.1 3.2 were out so it's not surprising that technological advancements have been made. ;) Then I came across cvsync: http://www.openbsd.org/cvsync.html Is cvsync preferred now? CVSup was able to either mirror the full repository, or make a checkout. The method you were looking at for CVSup was just for making a checkout. (This was quite widely used by FreeBSD in the past, but in OpenBSD the main method of users fetching the tree was from anoncvs mirrors). cvsync is only used for mirroring the full repo. Useful if connectivity between you and an anoncvs mirror isn't very fast, or if you want to hack offline and still be able to make diffs etc. Unlike CVSup it cannot do a direct checkout. I used to run a local cvsync mirror at home. But then the anoncvs server I used had some upgrades and got much faster so I now just fetch directly from there, unless I am going to be travelling and want an up-to-date local copy of the tree on my laptop. Now that I understand what cvsync is, I don't think it would have saved me any time with the updates as the longest time seems like my HDD searching for data, not the actual transmit. -- inum: 883510009027723 sip: jungleboo...@sip2sip.info xmpp: jungle-boo...@jit.si
Staying -current with cvsup or cvsync
Hello All, For the last several updates I've applied to my system, I've used plain CVS: cvs -q up -Pd This is pretty slow for some reason, but I understand that's just how CVS works. Michael W. Lucas' book Absolute OpenBSD (first edition) talks about using CVSup to update the local copy against the remote repo. (Page 344) I also found this page: http://www.openbsd.gr/cvsup.html (notice that this is NOT .org.) the .org site doesn't have the same page: http://www.openbsd.org/cvsup.html But the problem is I can't find cvsup in /usr/ports/net nor anywhere else: # make search key=cvsup # Does this mean cvssup is no longer used? Then I came across cvsync: http://www.openbsd.org/cvsync.html Is cvsync preferred now? If so, could you advise what to use for collections if you want to have the same effect of: cd /usr/src cvs -q up -Pd The example file displays: name openbsd release rcs But I don't know if that will yield the desired outcome. Thanks for any assistance. Best, j.b. -- inum: 883510009027723 sip: jungleboo...@sip2sip.info xmpp: jungle-boo...@jit.si
Re: Staying -current with cvsup or cvsync
Am 28.11.2014 21:33, schrieb Jungle Boogie:
Hello All,
For the last several updates I've applied to my system, I've used plain
CVS: cvs -q up -Pd
This is pretty slow for some reason, but I understand that's just how
CVS works.
Michael W. Lucas' book Absolute OpenBSD (first edition) talks about
using CVSup to update the local copy against the remote repo. (Page 344)
I also found this page: http://www.openbsd.gr/cvsup.html (notice that
this is NOT .org.)
On the footer of this site you will find
-- Quote --
This site Copyright © 1996-2009 OpenBSD.
$OpenBSD: index.html,v 1.605 2009/12/01 18:13:58 ajacoutot Exp $
-- end Quote --
so it's out of date and thus probably not authoritative.
the .org site doesn't have the same page:
http://www.openbsd.org/cvsup.html
But the problem is I can't find cvsup in /usr/ports/net
nor anywhere else:
# make search key=cvsup
#
Does this mean cvssup is no longer used?
Yes.
Then I came across cvsync: http://www.openbsd.org/cvsync.html
Is cvsync preferred now?
If so, could you advise what to use for collections if you want to have
the same effect of:
cd /usr/src
cvs -q up -Pd
The example file displays: name openbsd release rcs
But I don't know if that will yield the desired outcome.
Thanks for any assistance.
Best,
j.b.
The example file /usr/local/share/examples/cvsync/cvsync.conf
installed by the cvsync package also has
# # alternatively, fetch only selected parts
# collection {
# name openbsd-cvsroot release rcs
# }
# collection {
# name openbsd-ports release rcs
# }
# collection {
# name openbsd-src release rcs
# }
# collection {
# name openbsd-www release rcs
# }
# collection {
# name openbsd-xenocara release rcs
# }
#
# # the X11 and XF4 trees are of historical interest only
# collection {
# name openbsd-x11 release rcs
# }
# collection {
# name openbsd-xf4 release rcs
# }
so the third collection in this list
name openbsd-src release rcs
should be sufficient for you.
HTH
rru
Re: Staying -current with cvsup or cvsync
Dear Einfach,
From: Einfach Jemand rru@gmail.com
Sent: Fri, 28 Nov 2014 22:30:29 +0100
To: misc@openbsd.org
Subject: Re: Staying -current with cvsup or cvsync
On the footer of this site you will find
-- Quote --
This site Copyright © 1996-2009 OpenBSD.
$OpenBSD: index.html,v 1.605 2009/12/01 18:13:58 ajacoutot Exp $
-- end Quote --
so it's out of date and thus probably not authoritative.
Yes, I noticed that, too!
the .org site doesn't have the same page:
http://www.openbsd.org/cvsup.html
But the problem is I can't find cvsup in /usr/ports/net
nor anywhere else:
# make search key=cvsup
#
Does this mean cvssup is no longer used?
Yes.
Thank you for the confirmation, I'll disregard this section from the book and
see what the updated book (edition 2) has to say about updates.
Then I came across cvsync: http://www.openbsd.org/cvsync.html
The example file /usr/local/share/examples/cvsync/cvsync.conf
installed by the cvsync package also has
# # alternatively, fetch only selected parts
# collection {
# name openbsd-cvsroot release rcs
# }
# collection {
# name openbsd-ports release rcs
# }
# collection {
# name openbsd-src release rcs
# }
# collection {
# name openbsd-www release rcs
# }
# collection {
# name openbsd-xenocara release rcs
# }
#
# # the X11 and XF4 trees are of historical interest only
# collection {
# name openbsd-x11 release rcs
# }
# collection {
# name openbsd-xf4 release rcs
# }
so the third collection in this list
name openbsd-src release rcs
should be sufficient for you.
Ah, the example was very close to what I was trying. I tried many variations
to have src in the collection but it wouldn't work.
I'll give this a shot and see how much faster the update is with cvsync!
HTH
rru
Best,
j.b.
--
inum: 883510009027723
sip: jungleboo...@sip2sip.info
xmpp: jungle-boo...@jit.si
Re: Staying -current with cvsup or cvsync
Am 28.11.2014 22:38, schrieb Jungle Boogie: [...] I'll give this a shot and see how much faster the update is with cvsync! You are aware that this might not be much faster since - first you synchronize your local repository with cvsync, which takes some time - then you synchronize your working copy /usr/src with your local repository doing cvs up, which takes time as well. See also http://marc.info/?l=openbsd-miscm=138652306002368w=2 Cheers, rru
Re: Staying -current with cvsup or cvsync
Dear Einfach, From: Einfach Jemand rru@gmail.com Sent: Fri, 28 Nov 2014 22:59:05 +0100 To: misc@openbsd.org Subject: Re: Staying -current with cvsup or cvsync Am 28.11.2014 22:38, schrieb Jungle Boogie: [...] I'll give this a shot and see how much faster the update is with cvsync! You are aware that this might not be much faster since - first you synchronize your local repository with cvsync, which takes some time - then you synchronize your working copy /usr/src with your local repository doing cvs up, which takes time as well. See also http://marc.info/?l=openbsd-miscm=138652306002368w=2 This helped out! I'll stick with the traditional way. Cheers, rru -- inum: 883510009027723 sip: jungleboo...@sip2sip.info xmpp: jungle-boo...@jit.si
