Search results

Thanks for all the replies, this is really helpful! I didn't realise 2011-1 and 2011 were not the same socket, strangely enough, if you run the CPU comparison by Intel, it doesn't show that their sockets are different (both are listed as FCLGA-2011 and this row in the table is not highlighted)...

Hi all, I have a couple of dual ESXi servers running E5-2690v2 (Ivy Bridge, with 10 cores each up to 3.6GHz turbo) in 2 different X9 Supermicro boards (X3DRL-3F/ATX and X9DR3-F/E-ATX). I'm thinking on rejuvenating by dropping in some high-core count v3/v4 processors allegedly using 2011-R1...

@efschu2 , the bench results at the end, you have 3 ATTO graphs, what's the difference between them? Also, why did you insist on sync=always? This totally destroys write performance for me. Below are my graphs with sync=standard and sync=always for comparison.

Yes

Not at all. You can add multipathing the same way as for iSCSI. Have a look at VMware documentation for iSER and you'll see all these details there.

iSER initiator setup for VSphere 6.7 is very straightforward and doesn't deal with any particulars of RoCE, as it's all high level. Look up VMware documentation for this. What are you planning to use as your iSER target? I use SCST on CentOS (over ZFS) and, again, I don't need to deal with any...

Yes, I meant RoCE, and, indeed, CX-3 supports v1, but only CX-3 Pro and up support v2, which is quite a bit of a difference latency-wise. I have done some measurements with my setup using SRP a while ago, now need to redo the same baseline to see how things are different with iSER, but, from...

Guys, is there any reasonable (price-wise) substitute to CX-3 cards available now in the wild? My CX-3s are fine, but they don't perform with RDMA-over-IP as well as newer cards would, impacting iSER performance comparing to what SRP used to provide... Has to be at least 56Gbps in EN mode though...

I'm not aware of any overviews, but, AFAIK, if you want to stick with the recent versions of ESXi (6.5 and above), your only option is Linux with ZoL and iSER (I use SCST implementation, but the built in should work as well). Edit: and yes, it is slower than pre-6.5 SRP had been...

A remark on IPoIB vs EN mode: everything else being equal (FDR and 56GbpE), you get SIGNIFICANTLY less network throughput in IPoIB mode comparing to native EN. And when I say significantly, I mean 1/2 the throughput as the best case scenario. That said, as long as storage speed is the main...

I think point in case here is one shouldn't reset the env variables in UBoot, otherwise bad things happen. Can someone with a working switch send this fellow their output of printenv? I would do it myself, but travelling at the moment... Thanks!

The guide is available, no need to worry. Just PM me. As for the "right or wrong", just avoid buying the SX6XX5 models (which don't have a CPU) and you will be fine, as long as it's EMC. If it's Dell or HP, then it's even better, as you won't need to do the conversion.

Sorry mate, I'll do when I reboot my switch. I don't do this very often...

@Hindsight this is strange. The only case I had such a problem was when I tried putting CX-3 Pro firmware on my CX-3 (non-Pro) cards. Then they lost 56Gbps speed. But, otherwise, never had issues like this...

As @Hindsight pointed out, you need to run opensm first. Make sure it's configured properly. The lights will come on then. If anything else fails, you can also run flint on the switch itself. PM me for details.

The short answer is no, at least not easily. PM me for the longer answer...

You can't enable both SM and Gateway at the same time...

Glad it worked for you. Still not sure why it didn't work in the first place. FYI, I now (after the system has been setup) have the link broken and everything works fine...

I'd try "setenv rootdev=/dev/mtdblock7" instead of what you've been doing and give it another try. Also, after you changed the shell in /etc/passwd, are you getting a different prompt? Can you post the prompt you're getting?

So you're changing wrong files. When you boot into scratch, mount MLNX-OS partition (in RW mode!) and double check the files. In "virgin" state, /etc/passwd and /etc/shadow (the path are relative to MLNX-OS root!) are symbolic links to files on another partition (don't remember exact path). Make...