Ray Van Dolson (r...@bludgeon.org)

It gets even worse... whoever had set up the system first didn't now
how to get the IP address correctly from a variable in the ip-up
script.  So what'd they do?  They called grep on /var/log/messages to
look for it.

You can imagine the fun this created.... :-)

To the OP (sorry, jumping into a lot of threads late); what version of
vsftpd are you using?

A few months back FileZilla released a new version that "broke" TLS/SSL
support with a number of FTP servers.  I ran into the problem with
ProFTPD specifically:

http://bugs.proftpd.org/show_bug.cgi?id=3094

But vsftpd had this issue as well and was patched in v 2.0.7.  I don't
know if this fix was backported by RH or not....

I also don't know if SmartFTP client would exhibit the same problem.
You could try an older version of FileZilla (< 3.1.0) to see if it
works correctly...

Ray

The only thing I'll say in the world of using PPTP (via PoPToP) is to
consider what happens when most or all of your clients reconnect at one
time (network glitch, etc).  This was my biggest challenge as the
original configuration had PPP calling all sorts of perl scripts and
such from its ip-up mechanism.  The server would come to a complete
crawl as 800+ of these ip-up scripts would fire off along with their
associated tasks.  This would result in clients timing out, links
failing, etc -- the server could never "catch up".

The band-aid solution was to rate limit SYN packets that established
the connection... the permanent solution was to write a plugin for PPPd
in C that replaced most of the ip-up functionality with something a bit
more efficient.

As long as you're not needing to do any sort of complex post login
tasks for each user, this may not even end up being an issue.  But
something to keep in mind and plan for if you're talking 1500 users...
:)

Ray

Hmm, not sure.  He's from down under and works for HP (last time I
checked).

Ray

Yeah you'd definitely have to do some coding to make it manageable.
They always came in 'handy' but sounds like PPTP might be the way to go
for this particular question anyways.

PoPToP is rock solid in my experience and the maintainer is very
responsive and helpful (James Cameron).

Ray

How about lots of GRE tunnels? :-)

Ray

To OP; anecdotal evidence only -- and I certainly wouldn't recommend
using PPTP for a secure VPN solution :)  At my previous job we ran
PoPToP (PPTP) on CentOS and the older HP DL140 G1 1U servers and were
handling up to 1000 clients pretty comfortably per machine.  This was
with 1GB of RAM per server and a single 2.4GHz Xeon processor.

Left before we could migrate to OpenVPN which I think would have
slightly higher processing requirements. :)

Ray

I think that's why the SuperMicro riser I found was $150.  Others were
cheaper, but probably suspect.

I'm going to return my 12-way card and get two 8-way cards instead.

Thanks for the reply,
Ray

Fellow server-builders out there, this is for you. :)  I was trying to
build a cheap JBOD type storage solution running CentOS.  Ended up
snagging a Supermicro SC826TQ-R800LPB 2U case (12 drives slots) and a
Supermicro X7DBE-O motherboard.  Unfortunately, without thinking I
snagged a 3ware 9650SE-12ML SATA RAID card which is a full height card
and thus does not fit in my case.

I have a few options here:

  - Go with a 3U case instead.  Now I have a 16-drive case with only a
    12-way card. :-)
  - Return the 3Ware and order two of the 8-way cards which are
    half-height.  This means I love some of my space now however as I
    have to have an extra spare drive per card.
  - Get a PCI-E riser.

The riser option I like, except I also have the BBU unit for my RAID
card meaning I need to ensure that I have enough clearance and support
for it when the card is mounted horizontally.  You can mount the
battery elsewhere, but not the logic module...

Most of the PCI-E 8x risers I've seen out there seem to be pretty small
profile and I'm not sure if they'd give the clearance needed (see the
Supermicro provided CSE-RR1U-E8).  It looks like I need around 35mm of
clearance to get the card in sideways with the BBU attached...

Another solid option would looke to be a 2U riser, but I am having a
harder time finding somethign reasonably priced.  The RSC-R2UE-A3E8
from Supermicro looks perfect and would definitely give me the
clearance I need, but is priced 10x higher than the 1U riser!

Any advice?

Apologies for the off-topic post.. this *will* be running CentOS
however and there doesn't seem to be a more appropriate mailing list
for general stuff like this.. :-)

TIA,
Ray

Look inside the .spec file.  The Release line should contain a dist tag
as follows:

Release:        1%{?dist}

The "1" will vary of course.

Then, when you build, just define the dist flag to be what you like:

  % rpmbuild --define 'dist .el5' <whatever>.spec

Ray

Thanks for the reply John.  However, my question wasn't so much "if I
should" but how the xfs support in CentOS compares to jfs.  It seems to
me that xfs is a bit more up-to-date.

If you'd like, consider the question academic vs giving me a
recommendation that pushes me down the path of unsupported filesystem
doom. :-)

Ray

Hi folks... trying to pick between jfs and xfs for a filesystem.  In
the past we've used jfs with CentOS + centosplus, however, an older
post indicated that this may not be the best choice as the version of
jfs included with the centosplus kernel would only be as new as the
version that was included in the 2.6.18 kernel as RH doesn't backport
fixes...

It looks like xfs isn't part of the centosplus kernel, but instead is
provided as a kmod -- so I'm thinking it might be the better choice
based purely on the fact that it's likely to be current.  Is my
understanding correct there?

What would stop us from building a kmod-jfs against the latest jfs from
the up-upstream kernel and not building jfs.ko in the centosplus kernel
at all?  It looks like jfsutils is a fairly recent version...

Thanks,
Ray

First off; are there any storage-centric mailing lists that target
Linux specifically?

Secondly, I'm wondering if there's anything in the works or already out
there similar to Sun's L2ARC[1] for ZFS.  Basically this is a
filesystem (or maybe lower-level) cache for reads, writes that can be
comprised of faster media like SSD.  I think it could make SATA-based
JBOD solutions a lot more attractive from a performance perspective.

Anything out there along these lines already or suggestions on a better
place to dig for information?

Ray

[1] http://blogs.sun.com/brendan/entry/test