Hey, that's pretty good. I actually think I'll offer him one of my team as a dotted line report. He's the kind of person who would eat that up (his title is manager but he has no reporting employees). Also the guy who works for me wants to get into security and this would get him in deeper then if our sec dude saw him as the enemy. So would be good for his career as well. He's 25 and has the energy to keep up with his current workload as well, and having a reliable security infrastructure would make my team's work a lot easier.
what did you learn today?
- Thread starter chalex
- Start date
- Status
Because we don't need/want the entire size of the spindle, but are I/O bound and require the multiple spindles to acheive it. There are various reasons why you would not want all of the space.
Agreed, but if you assign the LUN to the exchange box, even if you don't mount it, it no longer looks like tempting free space to the uninformed. I typically write a file system and dump a readme in there just to make it obvious what's going on in this situation.
Do you guys not over commit your disks? I generally don't do it on the volume level (NetApp world, can have multiple LUNs in a volume) but LUNs and VMDK's are all thin provisioned unless it's a super HA "every thing else can crash first if needed" server.
Have not run into any issues yet, but it's weird to almost exclusively think of storage via IOPS and not space since that's how all our customers view it.
Have not run into any issues yet, but it's weird to almost exclusively think of storage via IOPS and not space since that's how all our customers view it.
Depends hugely on the application.
Bulk storage I overprovision with enthusiasm. Servers that collect data in real time from scanners used to image volunteers with experimental drug doses, where the I/O performance is critical get dedicated arrays. If you've not got/had disk I/O limited systems, you don't think of it as an issue.
I learnt that Sygate Firewall is a bear, especially when the user doesn't understand the popups. Even worse, the machine booted, and did not display any icon that said Sygate was running, nor did the actual management program work.
The symptom? Only DHCP and DNS worked. No TCP access at all. And this was after I removed a malware, so you can imagine what I was thinking.
Uninstalled Sygate, poof, everything works.
The symptom? Only DHCP and DNS worked. No TCP access at all. And this was after I removed a malware, so you can imagine what I was thinking.
Uninstalled Sygate, poof, everything works.
I detest software firewalls; I've had f-secure's firewall on a workstation pass the first 200GB of a 400GB backup, and then start blocking traffic causing the backup to fail.
I learned once again that vendors do not care once they have your dollars. Thanks for not checking on that package tracking # and leaving it up to us to find out that our power cables were lost by UPS and are going to put us behind by a few days.
I've learned that you shouldn't fix what was never broken.
Looking at you IBM. --
--
Your previous gen server rails were awesome.
These new ones... not so much.
Looking at you IBM. --
--Your previous gen server rails were awesome.
These new ones... not so much.
That people who have any combination of:
don't like to
don't know how to
don't think it is important to
perform testing will never see the light.
(Bug is encountered by beta tester of a new laptop deployment. Solution: hope it will not be widespread because "the tools to troubleshoot it do not exist". After I use the very existent tools and find the solution (a solution which adds roughly 60 seconds to each deployment since all of the other laptops were imaged before making sure the image was good) : "no, that's included in a service pack")
Yes, sometimes computers do not do what you expect them to. But testing can tell you many of the problems in advance.
don't like to
don't know how to
don't think it is important to
perform testing will never see the light.
(Bug is encountered by beta tester of a new laptop deployment. Solution: hope it will not be widespread because "the tools to troubleshoot it do not exist". After I use the very existent tools and find the solution (a solution which adds roughly 60 seconds to each deployment since all of the other laptops were imaged before making sure the image was good) : "no, that's included in a service pack")
Yes, sometimes computers do not do what you expect them to. But testing can tell you many of the problems in advance.
This could be said of HP as well, but not their rails (the G5 and G6 rails are gorgeous). Instead, on the G5, the cable management arms are solid metal, supported on both sides of the rack.
So, a G5 cable arm:
http://apcmag.com/images/serve...re/tour/DSC_0033.jpg
Nice and solid, doesn't flex under the weight of the SCSI cables, KVM cabling, network and power.
A G6 cable arm (can't find a pic) is only supported on one side, and even without cables bends down towards the ground.
Railing on rails...
We have been using really nice SMC cabinets with square holes. We use a lot of Dell Rapid rails. They are generally great, although each generation seems lighter and cheaper.
We are partnering with IBM on a new data center and they sent us some of their racks to look over. Narrow round-holes pieces of crap with no room for anything in them.
I wanted to order some Dell versa-rails to see how various stuff fits in them. The funny/awesome news is that the nine generation rapid-rails, or at least the 2950s, have these twirly ends on them that morph them from being perfect for square, to pretty decent for round. I saved more than a couple grand figuring that out before I ordered stuff I didn't need.
Honest...they twirl...who knew.
We have been using really nice SMC cabinets with square holes. We use a lot of Dell Rapid rails. They are generally great, although each generation seems lighter and cheaper.
We are partnering with IBM on a new data center and they sent us some of their racks to look over. Narrow round-holes pieces of crap with no room for anything in them.
I wanted to order some Dell versa-rails to see how various stuff fits in them. The funny/awesome news is that the nine generation rapid-rails, or at least the 2950s, have these twirly ends on them that morph them from being perfect for square, to pretty decent for round. I saved more than a couple grand figuring that out before I ordered stuff I didn't need.
Honest...they twirl...who knew.
The next generation of Dell rails (R410/610/710) don't do that rotation trick anymore. Instead they have pegs that are square at the base, and then round off, so if you have a rack with round holes, they go in halfway, and if you have square holes, they go in all the way.
I'm learning that sometimes, just sometimes, I enjoy being part of the proverbial problem--especially when dealing with people for whom I don't particularly care.
"Oooooh, hmm, no. No, I can't approve your solution and instead I'm going to call a formal design study for that. I just don't feel comfortable implementing your solution as it stands. No reflection on you, but the potential impact is too high. Also, before the design study can even have its first kickoff meeting, I'm gonna need a full business case, a standards deviation justification, and an open RFC. And opening the RFC will require senior management's approval."
"Oooooh, hmm, no. No, I can't approve your solution and instead I'm going to call a formal design study for that. I just don't feel comfortable implementing your solution as it stands. No reflection on you, but the potential impact is too high. Also, before the design study can even have its first kickoff meeting, I'm gonna need a full business case, a standards deviation justification, and an open RFC. And opening the RFC will require senior management's approval."
I hate the G6 rails for the DL360. They hardly slide. I feel like I should grease them they pull out so hard.
This iw why the BOFH episodes frequently have 'please degauss your monitor by rapidly flicking the power on and off' --
--Haha! I thought I was the only one who remembered the original BOFH. Don't forget to hook up the car battery to the data center doorknob before you go on summer vacation...
Well after a very exciting weekend, I've come to appreciate how far a full drive backup with Backup Exec 12.5 can take you when the primary drive on a RAID1 crashes and the spare has bad sectors.
Typically I know most people just backup their data and not the whole kitten kaboodle (Program Files, Windows, etc) because they figure if something were to happen they could just rebuild, reinstall, and restore. Well, in this case I was able to get a server restored onto a RAID5 of new disks in less than 3 hours (if you take out the time waiting on the drives to initialize and dinking with the PowerVault 124T for Windows to detect the drive and changer). This included OS reinstall, drivers and Windows Updates, Backup Exec install, and autoloader reconfiguration. I just told it to restore all of C and the System State, then once it was done rebooted the server and we were back in business. From there, it took restoring the Exchange mailboxes into the Recovery Storage group and merging the databases into the live mailboxes, and 48 hours after the server crashed all was well. Lucky this was on the weekend!
I couldnt have imagined having to tell the owner I needed to rebuild their AD domain and Exchange from scratch, I probably would have rather jumped off the 35W bridge.
Typically I know most people just backup their data and not the whole kitten kaboodle (Program Files, Windows, etc) because they figure if something were to happen they could just rebuild, reinstall, and restore. Well, in this case I was able to get a server restored onto a RAID5 of new disks in less than 3 hours (if you take out the time waiting on the drives to initialize and dinking with the PowerVault 124T for Windows to detect the drive and changer). This included OS reinstall, drivers and Windows Updates, Backup Exec install, and autoloader reconfiguration. I just told it to restore all of C and the System State, then once it was done rebooted the server and we were back in business. From there, it took restoring the Exchange mailboxes into the Recovery Storage group and merging the databases into the live mailboxes, and 48 hours after the server crashed all was well. Lucky this was on the weekend!
I couldnt have imagined having to tell the owner I needed to rebuild their AD domain and Exchange from scratch, I probably would have rather jumped off the 35W bridge.
Depending on the amount of data a setup /disasterrecovery and mailbox restore with logs off might have been much faster than 48 hours. I've done it in as few as 24 including the BE install, tape catalog from media and data restore.
I can play that tune in an hour. Requires: vm environment, netapp, snap manager for exchange, sql, snapmanager for VI and potentially another netapp at the dr site. Got the whole core infrastructure back as fast as I could install ESX/ESXi and point at the datastore. Good times.
I hate DR drills. Security guy actually whipped out a table in the meeting rolled some dice and said: data center 1 hit by a hurricane is the scenario. These systems are gone, these are unusable. It was like some kind of retarded random encounter in real life. Then we run around and prove we can recover from it.
Heh, I think you meant "kit and kaboodle". Also, why does your bridge have a small power supply? --
--Originally posted by ronelson:
Also, why does your bridge have a small power supply? --
I believe that is I-35 West.
A lesson learned again and again: KISS
Although it's neat to create a single disk deployment solution and just give each of the "Install Drivers" steps a WMI condition (like "Precision WoskStation T3400" or "VMWare etc"), as soon as you think you are getting tricky and throw in a few more WMI conditions the ANDs and the ORs get angry. And really, what did the extra conditions do for you?
Although it's neat to create a single disk deployment solution and just give each of the "Install Drivers" steps a WMI condition (like "Precision WoskStation T3400" or "VMWare etc"), as soon as you think you are getting tricky and throw in a few more WMI conditions the ANDs and the ORs get angry. And really, what did the extra conditions do for you?
Most of our customers are small businesses who don't see the value of DR (lucky this one forked out 5K to throw a tape library at their datacenter earlier this year so we could get a full backup) until it bites them on the ass. We're making a very strong push for virtualization across all of our customers because of the easy of management/backup/DR. It's a slow road but events like this make people open their eyes.
Cable contractors suck, that's why. We are having some fiber quoted and have 4 dramatically different prices for the same run. Same reason I got billed a couple years ago for 2000 feet of cable for 4 cat6 runs. When I asked how they got past the 300 feet limit of cat 6 (each less than 150 feet in reality), they said oops.
Today I learned that our main filesystem with 59M files has 50M files that are smaller than 256KB.
A FS-C5030N that prints an all-black page probably has the green cleaning arm incorrectly inserted.
That doing a data migration from an ancient ML350 G4P SATA storage server to replace the drives sucks.
4 x 250GB (RAID5) on an adaptec 2610SA raid controller, which supports only 48 bit addressing and a maximum single volume size of 2TB. The client bought 4 x 1GB SATA drives to replace, so right now I get to do that migration with little more than another server (RAID0, 850GB of storage - just what I really want) and 48 hours.
The server itself has ~1 million files @ 680GB.
4 x 250GB (RAID5) on an adaptec 2610SA raid controller, which supports only 48 bit addressing and a maximum single volume size of 2TB. The client bought 4 x 1GB SATA drives to replace, so right now I get to do that migration with little more than another server (RAID0, 850GB of storage - just what I really want) and 48 hours.
The server itself has ~1 million files @ 680GB.
We've put out bids for ISP's to provide us connections between various sites. The responses all involve them contracting out Telus to do the physical and virtual connection. Telus's bid was faaaar more expensive then all the others.
Wha?
We've had the same issue with Telus, connecting two sites w/ fibre. They were over double the competitor's pricing iirc, if not more.
- Status