<html><head></head><body><div dir="auto">We are using this volume for video editing, mainly it's video/audio/photo storage plus some project files replicated twice (previously 3 times). On a physical level single node contains 6x12 bay RAID0 enclosures joined into a logical volume. I've been playing earlier with a different configuration, like setting up a DRBD resource for each enclosure, then join it into a logical volume, but finally decided to go with a single volume due to simplicity despite some loss in replication speed. At this moment we're able to achieve about 1.2GB while replication which is little bit slower then a single enclosure speed, but enough for all our needs. Potentially we may grow up above PB next year or two. Do you advise any different setup?<br><br></div>
<div dir="auto"><!-- tmjah_g_1299s -->//OE<!-- tmjah_g_1299e --></div>
<div class="gmail_quote" >On 24 Jun 2019, at 20:17, Lars Ellenberg <<a href="mailto:lars.ellenberg@linbit.com" target="_blank">lars.ellenberg@linbit.com</a>> wrote:<blockquote class="gmail_quote" style="margin: 0pt 0pt 0pt 0.8ex; border-left: 1px solid rgb(204, 204, 204); padding-left: 1ex;">
<pre class="blue">On Fri, Jun 21, 2019 at 02:33:08PM +0200, Robert Altnoeder wrote:<br><blockquote class="gmail_quote" style="margin: 0pt 0pt 1ex 0.8ex; border-left: 1px solid #729fcf; padding-left: 1ex;"> On 6/21/19 9:16 AM, Oleksiy Evin wrote:<br><blockquote class="gmail_quote" style="margin: 0pt 0pt 1ex 0.8ex; border-left: 1px solid #ad7fa8; padding-left: 1ex;"> I've tried to update the kernel and DRBD 9 to the last available<br> version, but nothing helped with the PingAck issue. So I had to<br> downgrade to DRBD 8.4 which is started to replicate fine<br> [...]<br> Can anyone help me on the "PingAck did not arrive in time." repeated<br> error while initial bitmap synchronization?<br> [...]<br> I'm using the basic drbd configuration on 527TB LVM volume<br></blockquote> <br> This is not generally a version problem, except maybe for the<br> performance side effects of various versions. It's probably actually a<br> size problem. Bitmap operations on a 527 TB volume take some time to be<br> processed, and some operation probably blocks one of the DRBD kernel<br> threads long enough to cause DRBD to be too slow to answer a DRBD PING<br> request from its peer.<br> <br> There is an option "ping-timeout" (in the "net" section, the value being<br> in tenth of a second) that can be adjusted to change the ping timeout.<br></blockquote><br>And for various reasons, not least of them, performance considerations,<br>using a single DRBD volume of that size is most likely not what you want.<br>If you really mean it, it will likely require a number of deviations<br>from "default" settings to work reasonaly well.<br><br>Do you mind sharing with us what you actually want?<br>What do you mean to use that volume for?<br></pre></blockquote></div></body></html>