Aww jesvs....

[CletusJones]

This actually makes me feel not so bad anymore.

It's interesting to run a disk free check command and see it load up that 19TB drive.

I think their initial setup was a joke. There are way to many other options they had, but I don't think they understood the implications of creating that huge an array. If it's that big, it doesn't seem to have any sort of redundancy. None in fact, it seems.

yeah, whoever set that up needs to seriously get fired. there should have been NO WAY WHATSOEVER you could have formatted it if it was only a mounted device. you should have only had read/write access on a small portion of it, not full control over the whole 19tb.

 

[Syrup Beaver]

Yea. It's setup between their rendering farm and the animators and such each have access to it to send their files too.
Theoretically, unless I missed something, one of the animators could kill the SAN just as easily. But I think they set up access rights, at least they better have.

Sounds more like a NAS than a SAN then.

 

[Mean Mr. Mustard]

yeah, whoever set that up needs to seriously get fired. there should have been NO WAY WHATSOEVER you could have formatted it if it was only a mounted device. you should have only had read/write access on a small portion of it, not full control over the whole 19tb.

sounds to me like someone is using you to cover their ass.
scapegoat anyone?

 

[CletusJones]

60gb is hardcore. Yea...but I've learned from this that even the best computers and mainframe's, if set up shitty, are still pieces of crap.

unisys was here for 3 weeks setting it up - the right way. it's all mirrored and striped from one array to the other, and that's protected against any accidental deleting. the only thing that has any kind of access to the storage array on it is devices in the cabinet of the mainframe itself. it's a sweet setup.

since i work in banking, i should probably explain this a little better since not a lot of people work in my field...

all the terminals that process transactions feed a string of data to the TP (transaction processor) which are several processes that run on the mainframe, and the perform the transactions and change the data in the database fields required. it's pretty slick actually.

 

[Syrup Beaver]

unisys was here for 3 weeks setting it up - the right way. it's all mirrored and striped from one array to the other, and that's protected against any accidental deleting. the only thing that has any kind of access to the storage array on it is devices in the cabinet of the mainframe itself. it's a sweet setup.

since i work in banking, i should probably explain this a little better since not a lot of people work in my field...

all the terminals that process transactions feed a string of data to the TP (transaction processor) which are several processes that run on the mainframe, and the perform the transactions and change the data in the database fields required. it's pretty slick actually.

That's how all EMC Symmetrix's work, the only way to truly 'format' the system is from the service processor in the cabinet.

You could delete all of the meta's in the box from symcli or ECC, but it would take hours.

 

[wr3kt]

yeah, whoever set that up needs to seriously get fired. there should have been NO WAY WHATSOEVER you could have formatted it if it was only a mounted device. you should have only had read/write access on a small portion of it, not full control over the whole 19tb.

That's what's really irking me. I was almost positive that it only had read/write access to the mounted point. But it could see all of the 19TB.

 

[wr3kt]

That's how all EMC Symmetrix's work, the only way to truly 'format' the system is from the service processor in the cabinet.

You could delete all of the meta's in the box from symcli or ECC, but it would take hours.

Yea...if they gave the computer I was working on full rights to the SAN/NAS then the initial setup is a complete joke and hanging by a thread.
It seems like they have a really weak raid 5 or 0 setup in the mainframe with no sort of mirroring in case something like this happens.
Like I said, they were restoring the lost data from tape backup when I called. It was only a one day loss, but it's still really annoying that it happened even though it really shouldn't have.

 

[Syrup Beaver]

 

[CletusJones]

That's how all EMC Symmetrix's work, the only way to truly 'format' the system is from the service processor in the cabinet.

You could delete all of the meta's in the box from symcli or ECC, but it would take hours.

that's cool. i've never dealt with any kind of managed network storage like this before so it's all new to me on the little intricacies of it.

 

[CletusJones]

http://img8.exs.cx/img8/691/cnemc1.jpg[IMG][/QUOTE]
633 TB??? wow.

 

[wr3kt]

[IM]http://img8.exs.cx/img8/691/cnemc1.jpg[/IMG]

Sweet jesus!!!

 

[BigDov]

633.19tb raw capacity..... nice

 

[wr3kt]

That is one heavily mirrored setup, though.

 

[Syrup Beaver]

that's cool. i've never dealt with any kind of managed network storage like this before so it's all new to me on the little intricacies of it.

It takes a while to get used to, luckily I was given a test system to break to my hearts content long before we implemented them in production

 

[Syrup Beaver]

That is one heavily mirrored setup, though.

Yeah, that's all RAID 1 with striping accross multiple volumes accross multiple channels.

 

[wr3kt]

Yeah, that's all RAID 1 with striping accross multiple volumes accross multiple channels.

Damn...if this mainframe had any sort of mirroring, this wouldn't have been a problem.
I honestly think they setup a huge raid5 system or raid0 with the amount of storage they have versus raw storage.

 

[Syrup Beaver]

Damn...if this mainframe had any sort of mirroring, this wouldn't have been a problem.
I honestly think they setup a huge raid5 system or raid0 with the amount of storage they have versus raw storage.

Raid 5 isn't bad, it performs well enough on the high end systems.

We'll actually be trying it out on the next few DMX's that come in.

 

[wr3kt]

Raid 5 isn't bad, it performs well enough on the high end systems.

We'll actually be trying it out on the next few DMX's that come in.

Not when it's the ONLY raid system you're using with no sort of mirroring.

 

[wr3kt]

So I just got some news:
Apparently the system I was working on was directly attached to the SAN and the mainframe via some modules that were loaded synchrously with the other main SAN. And when those modules weren't loaded correctly, the mainframe went down about an hour after I left.
So in essence, anyone who would have worked on the system just as I had just restarting it would have knocked it out. I just happened to be working on it. And I can sense that if they don't resolve that issue, then whoever comes in and works on that raid array is going to take down their system again.
Jesus...

 

[Desslock]

So I just got some news:
Apparently the system I was working on was directly attached to the SAN and the mainframe via some modules that were loaded synchrously with the other main SAN. And when those modules weren't loaded correctly, the mainframe went down about an hour after I left.
So in essence, anyone who would have worked on the system just as I had just restarting it would have knocked it out. I just happened to be working on it. And I can sense that if they don't resolve that issue, then whoever comes in and works on that raid array is going to take down their system again.
Jesus...

Sounds like a real smooth operation.