Issues reading from my storage (Solaris, Napp-it & SMB)

Notice: Page may contain affiliate links for which we may earn a small commission through services like Amazon Affiliates or Skimlinks.

KeeZor

New Member
Oct 30, 2013
6
0
0
Hello STH users.

I have some frustrating issues reading from my storage drives.

My server is running Vmware Hypervisor 5.5, with a VM running Solaris 11.1 Cli, with napp-it installed on top.

The issue started after my VMware bootsector got corrupted, and i had to make a new VMware installation.
On import of my VM's they all showed sign of corruption, so i decided to make a clean start, and create new VM's instead of repairing the old ones.

on a fresh install of solaris + napp-it i have no issues importing my pools, and sharing them with SMB to my windows workstation.
Writing data to the storage from my workstation gives no problems either, but reading is almost impossible.

When starting a transfer it starts at the usual 110-120 MB/s, and continues with this speed for 5-10 seconds, then it halves the transferspeed every second, until it is down at xx KB/s, where the transfer is just useless.
Sometimes the transfer also starts at a very low speed, and it never rises in speed again.

The VM performance log shows no sign of the network, cpu, disk or memory being challenged at all.

i ran a DD Bench through napp-it and it shows 498,05 MB/s write and 603,77 MB/s read, with a 12,8GB testfile (the VM has 8 GB Memory atm, while resolving the issue)

all input is much welcomed, and because im not exactly a solaris expert, i might ask some otherwise self-explanatory questions.
 

KeeZor

New Member
Oct 30, 2013
6
0
0
I should also point out that the Vmware/Solaris/napp-it combination has been working like a charm for me, for 1-2 years, until i made this reinstall.
 

gea

Well-Known Member
Dec 31, 2010
3,141
1,182
113
DE
I should also point out that the Vmware/Solaris/napp-it combination has been working like a charm for me, for 1-2 years, until i made this reinstall.

I also had stability problems when using e1000 on a default OmniOS VM under ESXi 5.5
It works either when using the VMXnet3 vnic or when adding the following lines to /kernel/drv/e1000g.conf (reboot required)

#Disable TCP segmentation offload
tx_hcksum_enable=0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0;
lso_enable=0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0;

Maybe similar to this problem:
http://kb.vmware.com/selfservice/mi...nguage=en_US&cmd=displayKC&externalId=2058692
 

KeeZor

New Member
Oct 30, 2013
6
0
0
#3 I have now tried using the VMXnet3 instead of E1000 Vnic, but the result is that it doesn't connect with the adapter at all.
ifconfig -a gives no results with the VMXnet3 Vnic.

I need a little more help to try editing the config file.
i know i can open the file by typing "vi /kernel/drv/e1000g.conf" but when im inside the file, editing seems to behave very weird, and using the arrow keys to navigate is the only action that seems to work.

Can you push me along a little toward how to do the editing, there must be some commands needed, that is way outside of my knowledge.

Another question, are these network problems likely to be related to using ESXi 5.5 instead of 5.1?
 

gea

Well-Known Member
Dec 31, 2010
3,141
1,182
113
DE
#3 I have now tried using the VMXnet3 instead of E1000 Vnic, but the result is that it doesn't connect with the adapter at all.
ifconfig -a gives no results with the VMXnet3 Vnic.

I need a little more help to try editing the config file.
i know i can open the file by typing "vi /kernel/drv/e1000g.conf" but when im inside the file, editing seems to behave very weird, and using the arrow keys to navigate is the only action that seems to work.

Can you push me along a little toward how to do the editing, there must be some commands needed, that is way outside of my knowledge.

Another question, are these network problems likely to be related to using ESXi 5.5 instead of 5.1?
Yes, this is an Esxi 5.5 only problem.
But why did you use vi = very inhuman?

Easiest way:
Enable root access via ssh (napp-it menu services-ssh),
then connect Solaris from Windows via winSCP as user root
and browse and edit Solaris via winSCP.
 
Last edited:

KeeZor

New Member
Oct 30, 2013
6
0
0
#5
Thank you very much, that was an exceptionally easy method for file editing.

The issue has not been resolved by editing that file either.

I'm wondering if i should just make a new ESXi 5.1 install instead, and taking the lesson with me, to background-check more thorough next time before i update something in relation to my VM's.

If you have other quick suggestions, they are stille very welcome, as i will not install 5.1 until tomorrow, about 18 hours from now.
It is however 10 PM here, so i will soon have a long replytime :)
 

gea

Well-Known Member
Dec 31, 2010
3,141
1,182
113
DE
#5
Thank you very much, that was an exceptionally easy method for file editing.

The issue has not been resolved by editing that file either.

I'm wondering if i should just make a new ESXi 5.1 install instead, and taking the lesson with me, to background-check more thorough next time before i update something in relation to my VM's.

If you have other quick suggestions, they are stille very welcome, as i will not install 5.1 until tomorrow, about 18 hours from now.
It is however 10 PM here, so i will soon have a long replytime :)
you can try
- install vmware-tools
- add a vmxnet3 vnic to your vm
- disable e1000 vnic
- boot vm and recheck
 

mrkrad

Well-Known Member
Oct 13, 2012
1,244
52
48
I found out there was a e1000/e1000e problem with 5.1u1-5.5 on 2003-2012R2 machines, so its possible that it affects other o/s. for now it is vmxnet3 or bust.
 

KeeZor

New Member
Oct 30, 2013
6
0
0
you can try
- install vmware-tools
- add a vmxnet3 vnic to your vm
- disable e1000 vnic
- boot vm and recheck
I can not install vmware-tools on my solaris 11 Cli, or at least i dont know how :)

I have now thanks to you pointing me in the direction of the NIC being the issue, solved the issue for now, by making passthrough from my Intel NIC, and it works now with stable transfers.
It seems to be capped at 65-70 MB/s, but i will try to tweak up to the performance i'm used to from my setup.
 

KeeZor

New Member
Oct 30, 2013
6
0
0
I found out there was a e1000/e1000e problem with 5.1u1-5.5 on 2003-2012R2 machines, so its possible that it affects other o/s. for now it is vmxnet3 or bust.
So this means installing 5.1(not U1) might be the longterm solution, if i want to use the E1000 controller, or is the issue only in 5.5?