Re: [lopsa-tech] Converting from RHEL to CentOS

So we bundled our RHEL->CentOS migration with a simultaneous Satellite->Katello migration, but I think this is what we did just for RHEL->CentOS: yum -y remove rhnlib abrt-plugin-bugzilla redhat-release-notes* # drop in centos yum repo files, remove RHEL repo file here yum clean all yum -y install

Re: [lopsa-tech] filesystem feature flags

The only coordination I have seen has been when the backup software and filesystem are designed by the same company (i.e. GPFS and TSM). That said, this is sort of what POSIX compliance is all about - you might miss the more advanced features but at least the basic stuff is codified. Skylar On We

Re: [lopsa-tech] scripting: What do you use if not bash?

To keep our shell scripts relatively consistent and avoid code reinvention, I've written a suite of shell functions that ends up in a sourceable file in /usr/local/etc. It contains functions for cleaning up temporary files and directories, set PATH appropriately based on OS, output the date in ISO8

Re: [lopsa-tech] scripting: What do you use if not bash?

I'm amazed by this as well. A particular pet peeve of mine is a (for instance) Perl script that looks like this: $out = `cmd | grep foo | awk '{print $1}'| sed -e 's!baz!bar!g'` Not only is this something that can be done entirely in Perl, but it's something that can be done entirely in awk as we

Re: [lopsa-tech] scripting: What do you use if not bash?

For simple procedural stuff, I use bash. We're a Red Hat/Solaris shop so while I try to stick to POSIX shell, sometimes using the bash extensions really is useful. For more complex stuff (real variable scoping, real exception handling, complex data structures), I'll use Python. I used to use Perl,

Re: [lopsa-tech] backing up your VMs

I'll second the recommendation for TSM. We don't use TDP VE but I can definitely vouch for the scalability. We're use it to track ~1 billion onsite file versions and 5PB on tape, with an equivalent number offsite. Our daily backup volumes vary from 10TB all the way to 60TB. For a storage system of

Re: [lopsa-tech] Darwinian IT

Another way to look at this issue would be that backup frequency and retention should be orthogonal. When and how you run a backup shouldn't change the fact that the existing data should be retained for however long the original policy stated. Skylar On Wed, Sep 9, 2015 at 7:21 AM, John Stoffel

Re: [lopsa-tech] Debian equivalent of ks.cfg?

I believe you're looking for a preseed file: https://wiki.debian.org/DebianInstaller/Preseed Skylar On Fri, Sep 4, 2015 at 5:36 AM, Adam Moskowitz wrote: > When I boot an RHEL/CentOS system from DVD or ISO, I can grab the system > at the boot menu and set "inst.ks=" and have it use a ks.cfg >

Re: [lopsa-tech] Linux HPC clustering educational resources

The beowulf list is another great resource: http://www.beowulf.org/mailman/listinfo/beowulf There's a good mix of theoretical and practical people on the list. While I agree in principal about disabling ssh access to nodes, in practice it doesn't work. People really like to be able to login and

Re: [lopsa-tech] 3rd-party SFPs

We've been using Fiberstore.com: http://fiberstore.com/ They ship directly from China (but quickly!), but the ability to select exact vendor/model combinations is really good, and the price is definitely right. We haven't needed their customer support too much, but when we have it's been decent.

Re: [lopsa-tech] HPC Mailing Lists

The Beowulf list is where I would start: http://www.beowulf.org/mailman/listinfo/beowulf Lots of sharp people, both on the systems engineering and scientific sides. Skylar On Fri, Nov 21, 2014 at 12:41 PM, wrote: > Good day. > > Does anyone know of any HPC-related mailing lists? > > We're thi

Re: [lopsa-tech] 10G to the server

On 06/03/2014 06:03 PM, Robert Hajime Lanning wrote: > On 06/03/14 17:44, Skylar Thompson wrote: >> I'm not aware of any 10GBASE-T <-> SFP+. I think you might be thinking >> of SFP+ direct-attach (DA) cabling, where each end of the cable is a >> SFP+ transceiver.

Re: [lopsa-tech] 10G to the server

On 06/03/2014 05:04 PM, Shawn Wallbridge wrote: > I really like the Myricom’s because they are supported by everything, > but the Chelsio is faster in my experience. > > As for switches, you could in theory get 10GbaseT SPF+ gbics (I think > they make such a thing) so you don’t need new cards. I'

Re: [lopsa-tech] is OpenSSH susceptible to the Heartbleed bug?

On 05/13/2014 01:29 PM, Brandon Allbery wrote: > > On Tue, May 13, 2014 at 4:21 PM, David Nolan > mailto:no...@managedandmonitored.net>> > wrote: > > While SSH is not affected directly by the heartbleed bug, if you > have a server that was affected by the heartbleed bug there is some >

Re: [lopsa-tech] System power requirements

I've been similarly surprised by how efficient modern boxes are. That said, boxes with a large number of CPUs can have a wide range of power requirements. We have some 64-core boxes that idle at ~150W but hit ~800W under high CPU load. I've found HPL to be a useful tool to get CPUs stressed out, bu

Re: [lopsa-tech] experience w. "Advanced Clustering Technologies" as HPC vendor?

On 12/20/2013 02:40 PM, berg...@merctech.com wrote: > > Does anyone have experience with "Advanced Clustering Technologies" > as an HPC vendor? > > http://www.advancedclustering.com/ I've never bought their hardware, but their online HPL tool is pretty useful for getting a first-pass HPL.d

Re: [lopsa-tech] meta-data searching on medium+ filesystems

On 11/12/2013 04:02 PM, Loic Tortay wrote: > On 13/11/2013 00:24, berg...@merctech.com wrote: >> >> I've got about 45TB (wasn't that once considered 'large'?) of data on >> a GPFS filesystem. >> >> I'm looking for efficient ways to find files, based on metadata. >> >> Running "find / -ls" is not a

Re: [lopsa-tech] large scale storage - medium bandwidth

On 09/16/2013 02:47 PM, Lawrence K. Chen, P.Eng. wrote: > Wonder how old our LTO1's werewhen we upgraded our system to LTO3, one > unit that we did archives for didn't want initially have to pay for the > archive tapes...so we said they could have their archives written to the old > LTO's fo

Re: [lopsa-tech] large scale storage - medium bandwidth

On 09/15/2013 07:42 AM, Adam Levin wrote: > > Regarding protecting against corruption, that's the point of the erasure > coding. Mathematically, there can't be any, or at the very least, if > there *is*, it's corrected right away by calculations from the rest of > the data. The data is spread fa

Re: [lopsa-tech] Where do you buy parts for Sun servers?

-BEGIN PGP SIGNED MESSAGE- Hash: SHA1 On 08/20/2013 04:17 PM, Hugh Brown wrote: > Hi all -- I've got a Sun server that I need a power supply for, > and the only places to buy them I can seem to find are dodgy > websites with zero history. Has anyone got a vendor they can > recommend? We

Re: [lopsa-tech] Backup Reliability

On 05/21/2013 08:55 AM, Charles Polisher wrote: On Sun, May 12, 2013 at 05:01:39PM -0400, John Stoffel wrote: Edward> Focus on reliability. ;-) No, you want to focus on restorability of your data in face of disaster. It does no good at all to have a reliable backup system if you can't get th

Re: [lopsa-tech] Backup Reliability

On 05/13/2013 06:13 AM, Edward Ned Harvey (lopser) wrote: From: tech-boun...@lists.lopsa.org [mailto:tech-boun...@lists.lopsa.org] On Behalf Of Skylar Thompson Second, we depend LTO's data validation while data are being written to tape. I don't want to say "all devices," b

Re: [lopsa-tech] Backup Reliability

On 05/13/2013 05:59 AM, Edward Ned Harvey (lopser) wrote: >What I think/could/ work, though, is if checksumming filesystems like ZFS >could expose the checksum data to user applications (like backup clients), The reason that's not possible is because the ZFS checksums don't relate to the files

Re: [lopsa-tech] Backup Reliability

On 05/13/2013 05:50 AM, Edward Ned Harvey (lopser) wrote: From: Skylar Thompson [mailto:skylar.thomp...@gmail.com] I've been a TSM admin for years, so I admit I'm biased, but at scale I don't think there is much competition to TSM. The advantage of the progressive incremental ba

Re: [lopsa-tech] Backup Reliability

On 05/12/2013 01:14 PM, Jack Coats wrote: IBM every few years seems to feel the need to change the billing method just to keep the market guessing about if things are 'legal' or not. Over the years I went thru several conversions of billing methods. The one I liked was pay for a license,

Re: [lopsa-tech] Backup Reliability

On 05/12/2013 09:16 AM, Andrew Hume wrote: On May 12, 2013, at 6:20 AM, Edward Ned Harvey (lopser) wrote: Without checking the internet, and before you listen to other peoples' anecdotes or anything, I'd like to hear your gut feel, I want to know what your natural instinct is. What do you th

Re: [lopsa-tech] Backup Reliability

On 05/12/2013 10:21 AM, Jack Coats wrote: Professionally I have long liked IBMs TSM Storage Manager product with the Disaster Recovery option, but that is out of the price range for most. I've been a TSM admin for years, so I admit I'm biased, but at scale I don't think there is much compe

Re: [lopsa-tech] Backup Reliability

On 05/12/2013 10:26 AM, Michael Tiernan wrote: On Sun, May 12, 2013 at 10:34 AM, Skylar Thompson wrote: How do you define reliability? I think that that's a darned good question. Skylar's pair of points misses a key definition. As a guy from Keane that I used to work with said, no

Re: [lopsa-tech] Backup Reliability

On 05/12/2013 06:20 AM, Edward Ned Harvey (lopser) wrote: Without checking the internet, and before you listen to other peoples' anecdotes or anything, I'd like to hear your gut feel, I want to know what your natural instinct is. What do you think about the reliability of the following tools

Re: [lopsa-tech] bash question

On 05/01/2013 04:23 PM, Andrew Hume wrote: i am embarrassed to ask this, but i can't figure out what is going on. the problem is i am using bash as my login shell (of course, this is my problem, pure and simple, but i am trying to be modern and submit my fate to the feckless taste of GNU software

Re: [lopsa-tech] Version controlling permission sensitive files

On 04/22/2013 12:49 PM, Matthew Barr wrote: On Apr 22, 2013, at 3:35 PM, Bill Bogstad wrote: I've been waiting for at least a decade now for a Linux distribution to pick one of the CMs out there and just start supporting using said CM as the default way to manage their distribution. It see

Re: [lopsa-tech] OpenManage

On 04/05/2013 08:26 AM, Jon Dustin wrote: I've installed OMSA on Linux servers, then used some command-line scripts to parse status and send to another monitoring system. This way I'm not saddled with a "Dell-centric" monitoring console, yet still can see inside these servers. We did the sam

Re: [lopsa-tech] c13 to c14 Y cable supplier?

On 04/02/2013 01:04 PM, Doug Hughes wrote: stayonline.com also servertech.com to work better with servertech pdus, but stayonline for general purpose. We have a tone of them of various sizes. I'll second StayOnline. We've never used them for

Re: [lopsa-tech] Earbud recommendations

-BEGIN PGP SIGNED MESSAGE- Hash: SHA1 On 02/21/2013 08:01 PM, Michael Gurski wrote: > With all this discussion about earbuds, I feel I should pass along > a (possibly overcautious) warning from the Safety person where I > work. I'd off-handedly mentioned that I'd just use my > noise-cancel

Re: [lopsa-tech] nfs hard links

Indeed, according to the NFSv3 RFC (RFC1813), the NFS server should set the filesystem attribute to FSF_LINK to true if the filesystem supports hard links. The server should return NFS3ERR_NOTSUP if an attempt to make a hard link on a platform that does not support it. The NFSv4 has some this as a

Re: [lopsa-tech] linux and networking

-BEGIN PGP SIGNED MESSAGE- Hash: SHA1 On 10/17/2012 05:22 PM, Andrew Hume wrote: > screwed by linux again. sigh. > > so apparently i am overloading my pathetic linux system with too > much tcp/ip traffic. is there any way to detect this while (or > before or after) it is happening? of cou

Re: [lopsa-tech] linux and swapping

On 08/28/12 06:59, Brandon Allbery wrote: > On Tue, Aug 28, 2012 at 12:31 AM, Skylar Thompson > mailto:skylar.thomp...@gmail.com>> wrote: > > It can be a real problem for latency-sensitive applications that are > cohabiting on a system that's also doing heavy I/O.

Re: [lopsa-tech] linux and swapping

-BEGIN PGP SIGNED MESSAGE- Hash: SHA1 On 08/27/2012 09:25 PM, Brandon Allbery wrote: > On Tue, Aug 28, 2012 at 12:20 AM, Yves Dorfsman > wrote: > >> I for one, totally disagree with your statement, I do not want >> any of my process to be pushed to swap to just buy some buffer >> space.

Re: [lopsa-tech] linux disk device question

-BEGIN PGP SIGNED MESSAGE- Hash: SHA1 On 07/20/2012 01:00 PM, Lawrence K. Chen, P.Eng. wrote: > > > - Original Message - >>> From: da...@lang.hm [mailto:da...@lang.hm] Sent: Wednesday, >>> July 18, 2012 7:25 PM >>> Wait ... The firewall guys won't allow you to mount some >>

Re: [lopsa-tech] linux disk device question

On 07/18/12 03:52, Edward Ned Harvey wrote: >> From: tech-boun...@lists.lopsa.org [mailto:tech-boun...@lists.lopsa.org] >> On Behalf Of Andrew Hume >> >> i have two linux servers each of which has the same piece of SAN >> attache dto it as a LUN. that is, svra:/dev/sdbd is the same volume >> (or mo

Re: [lopsa-tech] Job/Process Profiling

-BEGIN PGP SIGNED MESSAGE- Hash: SHA1 On 06/19/2012 05:34 AM, Christopher R Webber wrote: > All, > > I have students and postdocs that write software as part of their > research and occasionally they want to answer questions like: - How > much memory does this take? - How much time does t

Re: [lopsa-tech] How to problematically identify the root volume.

On 04/13/12 06:20, Jonathan wrote: > Hi, > > stat --format='%D' / > > will give you the major & minor device of the root file system, if > that is what you need. > > Jonathan Oops, indeed. I misread the original email and thought Michael was looking for a way to programatically find the root files

Re: [lopsa-tech] How to problematically identify the root volume.

On 04/12/12 08:18, Michael C Tiernan wrote: > I am working on a small project and I have the need to identify the root > volume on a running linux system from inside a script. For the moment I'm in > the Red Hat EL environment but I'm expecting it to develop into a wider > application. > > I can

Re: [lopsa-tech] 120v single-phase, 208v split-phase and three-phase in the data center.

On 02/27/2012 04:49 PM, Skylar Thompson wrote: > On 02/27/2012 04:40 PM, Edward Ned Harvey wrote: >>> From: tech-boun...@lists.lopsa.org [mailto:tech-boun...@lists.lopsa.org] >>> On Behalf Of Luke S. Crawford >>> >>> On Mon, Feb 27, 2012 at 04:02:44PM -0800

Re: [lopsa-tech] 120v single-phase, 208v split-phase and three-phase in the data center.

are each 120deg out of > phase with each other. > > 120V = 1 phase > 240V = 2 phase > 208V = 3 phase It all depends on how you measure the voltage. 240V split-phase is 120V phase-to-ground and 240V phase-to-phase. 208V can come as single-phase, and is 208V phase-to-phase an

Re: [lopsa-tech] Noise levels in a workspace shared with servers?

On Thu, Feb 2, 2012 at 11:34 AM, Zack Williams wrote: > Skipping the whole "don't put people and servers together" discussion... > > Does anyone know where to find guidelines, either practical or regulatory > (I'm thinking OSHA here) for noise levels, exposure time, etc. in a space > that is used

Re: [lopsa-tech] DHCP options in PXE boot agents?

On 01/03/2012 10:39 AM, James Braid wrote: > On 26 December 2011 18:21, Skylar Thompson wrote: >> Unfortunately, all the PXE boot agents I've looked at don't support >> supplying custom options, which means we have to open up our DHCP server >> to accept packets f

[lopsa-tech] DHCP options in PXE boot agents?

boot agent. Thanks in advance for any suggestions, - -- - -- - -- Skylar Thompson (sky...@cs.earlham.edu) - -- http://www.cs.earlham.edu/~skylar/ -BEGIN PGP SIGNATURE- Version: GnuPG v1.4.11 (GNU/Linux) Comment: Using GnuPG with Mozilla - http:/

Re: [lopsa-tech] keystroke loggers than work in a chroot?

of the chroot kind of moot -- > proc, pts, utmp, etc. > > Anyone know of a simpler keystroke logger? We don't need to > sudo-related bits, just the command logging as the user themselves > functionality. > Could you use something like pam_tty_audit for this? -- -- Sk

Re: [lopsa-tech] NAS Recommendations

-sized directories that we can easily divide between two daily backup schedules, and assign one schedule per backup client. We've sized the backup clients such that either one could complete both schedules in under 24 hours. We've been using this backup methodology for about six mo

Re: [lopsa-tech] infiniband...

use Mellanox's OFED stack. One of the great things about industry "standards" is that everyone seems to have a different idea of how they should work. That said, though, we've never had any problems with Mellanox OFED. - -- - -- - -- Skylar Thompson (sky...@cs.earlham.edu) - --

Re: [lopsa-tech] infiniband...

We're also using it to run some of our large-scale backup operations using IPoIB. Before 10GbE came down in price and became better standardized, IB was a pretty cost effective way of backing data up. Without a lot of work or money, we could support streaming to multiple FC4-connected LTO

Re: [lopsa-tech] infiniband...

Running four streams dropped the aggregate down to 15Gbps. In fact, it appeared that either the HCA driver or the TCP stack is single threaded as top reported that ksoftirqd was pegging one CPU. I suspect either interrupts from the driver or processing TCP packets was responsible for that kernel t

Re: [lopsa-tech] Server Recommendations

-BEGIN PGP SIGNED MESSAGE- Hash: SHA1 On 1/15/2011 6:44 AM, Edward Ned Harvey wrote: >> From: tech-boun...@lists.lopsa.org [mailto:tech-boun...@lists.lopsa.org] >> On Behalf Of Skylar Thompson >> >> you can boot into UEFI and apply the updates there. >>

Re: [lopsa-tech] Server Recommendations

ell servers is that it's finally become possible to do firmware updates. You still can't do them in Solaris but you can boot into UEFI and apply the updates there. It's very handy not having to wrestle bootable USB sticks anymore. - -- - -- Skylar Thompson (sky...@cs.earl

Re: [lopsa-tech] Server Recommendations

L (and maybe SLES too). A lot of the DRAC/BMC functionality is configurable using IPMI, but some does require installing OpenManage or a vendor-specific tool (like MegaCli). The serial console can be redirect over IPMI too, so using IE isn't a requirement. - -- - -- - -- Skylar Thompson (sky...

Re: [lopsa-tech] Fwd: How to quantify amount of "churn" in a file system?

ll... One of the things that's been on my list of cool projects is to keep a log of changed files using inotify for backup purposes. Obviously it's Linux dependent, but it would still be useful not to have to scan some of our filesystems with 100 million files on them every day. - -- -

Re: [lopsa-tech] RAID cards with flash-backed cache?

> http://tools.rapidsoft.de/perc/perc-cheat-sheet.html > http://tools.rapidsoft.de/perc/perc-cheat-sheet.pdf Also of note - my memory is that the option names for MegaCli had slightly different spellings and capitalizations between versions. All in all, a real PITA to manage. - -- - -- Skylar T

Re: [lopsa-tech] RAID cards with flash-backed cache?

On 12/19/10 15:00, Skylar Thompson wrote: > Dell's PERC H700 and H800 cards support (but don't require) NVRAM cache. > LSI makes the silicon for the cards but I don't know that you can buy > them direct. We've only had them for a few months but we've had no >

Re: [lopsa-tech] RAID cards with flash-backed cache?

have issue with the volume of writes that occur to it. The one problem would be the batteries only last a few days without power. If you have an extended power outage, you'd want to be very sure you cleanly shutdown all your servers. - -- - -- Skylar Thompson (sky...@cs.earlham.edu) - --

Re: [lopsa-tech] RAID cards with flash-backed cache?

7;m wondering, though, is what success other people have had with > these cards, and with what Linux kernels? > Dell's PERC H700 and H800 cards support (but don't require) NVRAM cache. LSI makes the silicon for the cards but I don't know that you can buy them direct. We've

Re: [lopsa-tech] Faster than 1G Ether... ESX to ZFS

10GE, and you get > TONS of bandwidth at low latency. It also comes native with IP over IB > if you want to do that (L2) or you can go to the native IB storage > protocols as some have done (ISER/SRP), but that's even more outside. ESXi has pretty limited IB support, and even more l