ESXi 5.1 getting IO error uploading large files

[TdisalvoOrinoco]

Hello all,

I have noticed that with ESXi 5.1 I am getting an IO error whenever I am trying to upload a file greater than 1.3Gb.  I have tried to upload a file using the upload featuer within the Datastore browser.  I have also tried using FastSCP with no luck.

I looked up this problem and it looks like this was an issue back in 3.5 and they released a patch.  This is on a new build of ESXi 5.1 on an HP Proliant GL380 G6.  I am going from local disk on my laptop to a VMFS5  file store on DAS setup are RAID5 LUN.  If I can't get my ISOs up to the file storage, I will not be able to build any VMs.

Any help is really apreciated.  Bellow is the screenshot of the error that Veeam Backup and Replicate gives me when I try and do a SCP copy.

[onoski]

Have you tried checking veeams web site for this error or issue? If not it might be worth your while.

[TdisalvoOrinoco]

Ok, so I figured out why this is happening.  My log files were not stored on persistant disk, and apperently they were running out of space.  So when I was coping a large file this was causing the system to stop repsonding during the copy.  I think that this sounds odd, but it is what happened.  When I changed the location of my scratch files the copy slowed down a lot, due to the fact that the writes for the log file were on the same LUN as the writes for the data, but it did work.

So now I looked up why it did not create scratch stroage on persistant disk.  According to http://kb.vmware.com/selfservice/microsites/search.do?language=en_US&cmd=displayKC&externalId=103369...

Here are two examples where scratch space may not be automatically  defined on persistent storage. In each case, the temporary scratch  location will be configured on a ramdisk:

1. ESXi deployed on a Flash or SD device, including a USB key. Scratch  partitions are not created on Flash or SD storage devices even if  connected during install, due to the potentially limited read/write  cycles available.
   
   
2. ESXi deployed in a Boot from SAN configuration or to a SAS  device. A Boot from SAN or SAS LUN is considered Remote, and could  potentially be shared among multiple ESXi hosts. Remote devices are not  used for scratch to avoid collisions between multiple ESXi hosts.

So my question now is can I store the log files on the Flash drive, or is this a bad plan due to the amout of write operations that will take place?  The flash dirve made this system scream, so I may consider adding another piece of flash ram just for the log files.