what did you learn today?

Jump to latest Follow Reply
Status
Not open for further replies.
Muted (0)
G

Graeme K

Ars Legatus Legionis
14,776
Subscriptor++
  • #441
    Originally posted by Rob Banzai:
    Today I learned just how much crap we had in our storage room when I moved it all to the loading dock for the ewaste pick up. There's stuff in that pile from the 1980s. -- :mad: --

    -- View image here: http://i161.photobucket.com/albums/t238/rbanzai/ars/ewaste2.jpg --
    Click to expand...


    Yep. Did this last August. Boxed software on floppies. Sigh.
     
    R

    ronelson

    Ars Legatus Legionis
    21,399
    Subscriptor
  • #444
    Because if you leave it set to UTC time, it will reboot nearly every workstation on your network at 1pm in the afternoon.
    Click to expand...
    I wish everyone just worked off UTC times and there was no daylight savings crap. Then I could change my signature to "Shift: M-F 1100-2000 UTC" and never worry about people calling me at 5pm eastern because it is only 2pm there. I would just have to worry about people calling me at 5pm eastern because they are assholes who refuse to read my signature -- :) --
     
    S

    syndicate

    Ars Tribunus Militum
    2,280
  • #452
    Mostly I think it's been a project management disaster. Our VAR doesn't know much about what's going on, I don't think.

    We've been migrating all of our stuff fairly successfully over to NetApp, but since backup methods are changing (BCVs to Snapshots), our lower level environments are moving to FlexClones, it's difficult and nobody seems to have any clue *how* to do it.

    Nobody knows what the difference is between fractional reserve and space reservation and why you can't put multiple things in a single volume that you want to revert a snapshot on.

    We moved our VMWare environment to NetApp/NFS and every time we had issues with network communication our VMs were getting corrupted because multiple hosts tried to bring up the volumes. NetApp says to fix it with some NFS locking, but VMWare says that's not supported (something along those lines), so we moved it back to EMC SAN.

    Finally, there's SnapManager for Oracle... Nobody seems to know anything about this. Yet, every single oracle installation we have is on AIX and will eventually be using Veritas Storage foundation for Oracle RAC; apparently not supported by NetApp's SnapDrive (the underlying technology behind SMO). So we're kind of screwed there and if there's no solution for this -- I have no idea what we'll do. Not to mention the awesomeness that is the limitation of 128 luns per igroup and 500 volumes per filer(?) and some small number of total LUNs.

    I'm just a peon so I probably don't see the whole picture. But I have near zero confidence in the product, nor the way it's been handled.
     
    I

    Incarnate

    Ars Tribunus Angusticlavius
    9,004
    Subscriptor++
  • #456
    Originally posted by syndicate:
    We moved our VMWare environment to NetApp/NFS and every time we had issues with network communication our VMs were getting corrupted because multiple hosts tried to bring up the volumes.
    Click to expand...
    Of course systems don't like when they lose access to their storage. This is poor planning. I would always recomend a separate physical network whenever possible for this, or at a minimum a separate VLAN. Doesn't matter if it was FC, iSCSI, or NFS.

    But I have near zero confidence in the product, nor the way it's been handled.
    Click to expand...
    I don't think its the product, I think it's whoever is helping you guys do the implementation. Then again, maybe the vendor brought up better ways to do things, but other people didn't want to spend the money to do it right.
     
    S

    syndicate

    Ars Tribunus Militum
    2,280
  • #457
    Originally posted by Incarnate:
    Originally posted by syndicate:
    We moved our VMWare environment to NetApp/NFS and every time we had issues with network communication our VMs were getting corrupted because multiple hosts tried to bring up the volumes.
    Click to expand...
    Of course systems don't like when they lose access to their storage. This is poor planning. I would always recomend a separate physical network whenever possible for this, or at a minimum a separate VLAN. Doesn't matter if it was FC, iSCSI, or NFS.
    Click to expand...

    Agreed -- but one vendor (VMWare) recommends one thing and NetApp recommends something contradictory. I haven't been a party to that particular mess, but I do know that it's on FCP now.

    But I have near zero confidence in the product, nor the way it's been handled.
    Click to expand...
    I don't think its the product, I think it's whoever is helping you guys do the implementation. Then again, maybe the vendor brought up better ways to do things, but other people didn't want to spend the money to do it right.
    Click to expand...

    Part of it is the product as well. There are crazy limitations like number of volumes per filer, number of luns per igroup, etc. But I think the main thing is that the management web interface is garbage. We're going from a DMX to something that seems bush league in terms of FC. NFS is great don't get me wrong, but the FC management on NetApp is strange.
     
    pokrface

    pokrface

    Senior Technology Editor
    21,557
    Ars Staff
  • #458
    Originally posted by syndicate:
    We're going from a DMX to something that seems bush league in terms of FC. NFS is great don't get me wrong, but the FC management on NetApp is strange.
    Click to expand...
    That's because NetApp *is* bush league in terms of Fibre Channel, compared to competing offerings from EMC, HDS, and HP. You want NAS? NetApp murders just about everything in the field, and make EMC's Celerra look like a sick joke. You want FC? EMC's Symm DMX destroys NA's offerings. If you can afford it, DMX is one of the finest FC offerings in existence.

    That is in fact why NA pushes ESX on NFS so hard--it plays directly to their strengths, and to the advantages they have over all their competitors in general and EMC in particular. However, if you have the budget, there's basically no reason at all to go from ESX on FC to ESX on NFS. At best, ESX on NFS approximates some of the good things about ESX on FC at a substantial reduction in cost.
     
    N

    Neolithic

    Ars Praefectus
    4,904
    Subscriptor++
  • #462
    I learned it's important to test your workload on a new server before extolling its virtues.

    We are planning to migrate our production SAP server from a SunFire V880 to a SPARC Enterprise T5240.

    There are a couple of problems. First, SAP R/3 isn't a multithreaded app. Second, we have a relatively low volume of SAP use. The 8 year old design 1.0 GHz UltraSPARC III+ has better single thread performance than the 1 year old design 1.2 GHz UltraSPARC T2+.
     
    S

    shmooster

    Wise, Aged Ars Veteran
    186
  • #467
    Originally posted by Neolithic:


    We are planning to migrate our production SAP server from a SunFire V880 to a SPARC Enterprise T5240.
    Click to expand...

    Did you ask Sun about this? They no longer recommend Niagara servers for SAP due to the single thread performance. Found out this the hard way. X86 canes sparc on single thread anyway. If you must stay sparc look at the M series not the T series.
     
    V

    Vinnie Pasetta

    Ars Scholae Palatinae
    746
  • #468
    I learned that the lights flickering on and off in my office when followed a second or two later by someone shouting "we got a fire" gets your attention quick.

    The bad news is the 200 amp panel feeding the servers caught fire and melted down into a heap of metal.

    The good news the UPS held for the three hours it took to put out the fire and to run temporary power from another service panel.

    More bad news is nobody has power in their office and it will take two days to restore. (Guess who is working Saturday.)
     
    Megalodon

    Megalodon

    Ars Legatus Legionis
    36,693
    Subscriptor
  • #469
    I don't do a huge amount of desktop support but I help out sometimes. Crossloop is immensely helpful for helping out remote workers.
    Originally posted by Vinnie Pasetta:
    I learned that the lights flickering on and off in my office when followed a second or two later by someone shouting "we got a fire" gets your attention quick.
    Click to expand...
    lol...

    I live downtown, and there's a substation across the street from me, 4 big transformers. I was home sick one day:

    *lights flicker*

    "Hm?"

    *THOOOOOOOOM*

    I go over to the window and one of the transformers is on fire.

    Man oh man is the fire department not anxious to get anywhere near those things.
     
    V

    Vinnie Pasetta

    Ars Scholae Palatinae
    746
  • #473
    The main lug was not tight and when the coffee pot went bad, it somehow overloaded the main lug and caught fire.

    Here is the panel removed. Notice the missing lug.

    img0019y.jpg

    img0018u.jpg

    img0020stb.jpg
     
    B

    BajanDude

    Ars Praefectus
    4,468
  • #477
    Today I am learning that one of our Lenovo laptops is somehow missing about 10 updates, despite being managed by WSUS. This pisses SP3 off, and it takes 4+ hours to realise that it can't install. And adding those 10 updates manually doesn't populate $hf_mig$ properly either.
    Click to expand...
    Standalone SP3 works. Go figure.

    @Vinnie:

    Heh, small rack, no wonder you got such a long run time. We have several tonnes of battery in our server room, and we might get 15 - 30 minutes out of it. Mind you, we are powering 100 servers.
     
    Whittey

    Whittey

    Ars Tribunus Militum
    2,006
  • #479
    Originally posted by ronelson:
    Scroll down...

    http://nwtech08.blogspot.com/
    Click to expand...
    You mean this one? Not too shabby, especially for the small amount of equipment powering it. That is nothing compared to the "Day 1 in the data center" thread, but yours includes more fire -- ;) --
    Click to expand...
    Vinnie, Can you log into that weather server there and give me light rain from about 1am to 5am, then nice throughout the day? I've got some lawn patch that needs growing.


    -=Whittey=-
     
    V

    Vinnie Pasetta

    Ars Scholae Palatinae
    746
  • #480
    Originally posted by ronelson:
    Scroll down...

    http://nwtech08.blogspot.com/
    Click to expand...
    You mean this one? Not too shabby, especially for the small amount of equipment powering it. That is nothing compared to the "Day 1 in the data center" thread, but yours includes more fire -- ;) --
    Click to expand...


    Yes, that is it. I have another rack at a school which will get pulled over this summer. I have a small amount compared to many but I like it and by being a department of one, there is nobody else touching my stuff. -- :) --

    Originally posted by Whittey:
    Vinnie, Can you log into that weather server there and give me light rain from about 1am to 5am, then nice throughout the day? I've got some lawn patch that needs growing.
    -=Whittey=-
    Click to expand...


    Sorry, I use the weather server to make snow so I can enjoy snow days off from school. hehe
     
    Status
    Not open for further replies.
    Jump to latest Follow Reply
    Jump to latest Follow Reply