Search results

I am using the version that is linked here: https://forums.servethehome.com/index.php?threads/fusion-io-iodrive-2-1-2tb-reference-page.11287/page-21#post-302946

This is not just possible, but was done in production by the likes of Google many years ago. At the very first BSDCon somewhere around 1999 near Berkeley, the core team took a field trip to a nearby Globalcenter NOC to see Yahoo's cages. One of the founders of Yahoo had sponsored the...

Are there are no vsphere 7 drivers for the iodrive2? (couldn't find any, just want to confirm)

Many common RDMA protocols assume a lossless network, and will lock up or degrade dramatically if there is any RDMA packet loss. One way to detect the issue is to run a bandwitdh test such as ib_send_bw -a, that ramps up from low to high bandwidth. If it ramps up to %95 and stays there, great...

Hey @okrasit, I'm interested in the patches. I'll have spare D4040 soon, and I'd be glad to share whatever I do. Thanks!

Znapzend supports pre / post commands: oetiker/znapzend

I wonder, what is the most cost-effective kind of zil/slog device for high availability failover? I'm not yet doing HA, partly because by using kexec, reboots take around 20 seconds, which seems to be relatively travel free for both iscsi and NFS esxi clients. I'm interested in pursuing this...

If the filesystem supports it, it would not hurt to use a 4K block size. However, since most file systems use 4K or larger blocks anyway, the performance difference may be limited unless you need to tune it for something specific, like databases, where larger blocksizes are definitely an...

One option to troubleshoot this kind of issue is to use the mellanox bandwidth utilities to test increasing message sizes, to test whether its' a network issue. A lack of flow control will show up as a drop in bandwidth with increasing message size.

Wow, nice. I agree about netplan. jumbo frames in particular are difficult to configure using netplan. The bandwidth tests show very smooth curves, which looks very encouraging to me. I take it as an indication that flow control is working very consistently.

@Patrick, might you consider @gea involved in this? That would be a dream team! He's such a wealth of ZFS tuning experience.

Kudos to you for supporting a great cause! I donated.

Thanks for the confirmation. I went ahead and physically reversed the fans, and it went smoothly -- no issues, The sensors report the same speed, so everything looks good. Much appreciated. On a separate note, afterward I tried upgrading to 4.22.0F, and both 'show env cooling' and 'show env...

Is it possible to reverse the airflow for the fans on a Arista DCS-7050QX? Looking at the datasheet, there is an active circuit that uses a 3V PWM signal to control speed. Based on that, it seems that swapping fan leads will not work...

Ah, yea, sorry I was confused about whether esxi allowed values between 1500 and 9000. It does after all. I'll set up an ubuntu 18.04 host with scst or tgt to try to be able to compare results. What do you prefer? SCST or tgt? Here's how I setup a LIO target for esxi, for comparison...

Sorry for the delay. That's interesting that downgrading OFED improved SCST target support. Which flavor of linux is that? I might try that when I get a chance. Looking at your esxi logs, there are some login failures. If a CHAP user/pass are not set, I wouldn't have expected the login...

I asked the seller, "Is the original SMART data preserved, or has it been wiped? I'm interested in multiple if the SMART data has not been altered." He replied, "Unfortunately, this data is generally erased."

If they wiped the smart data to hide the fact that the SSD is worn out, and are offering it as "used" which ebay defines as fully functional., that's deceptive.

iser should be compatible between lio or scst target on ubuntu and vsphere initiator. I recall you may have said you are using multiple target addresses (or ports) for multipathing, so that shouldn't be the cause. What I do is watch the iser related messages such as: tail -f...

This may help both in terms of time to copy, and take less disk space. If the vmdk is 'Thin provisioned', then the 'flat' vmdk files is sparse, meaning there are many sectors that are unallocated. For thin provisioned disks, DD will expand the file. Other tools such as "cp --sparse=always"...