While keeping up with news from our industry, it caught my attention that among EPEL (Extra Packages for Enterprise Linux) users, servers running Rocky Linux have overtaken the others (RHEL, AlmaLinux, OracleLinux). While that may be an interesting fact, it might generate the impression that Rocky is more relevant than RHEL.
However, this notion requires further consideration. I am contemplating that RHEL users use RHEL for their critical workload. They need support coverage for the entire system from Red Hat, which rules out installing EPEL packages because Red Hat support does not cover them.
On the other hand, some developers need to have enterprise Linux. It is more convenient for them to use Rocky Linux or Alma Linux than to set up a free developer subscription with Red Hat. I think this fact proves only the point that developers prefer the RHEL clones over the original. (Devs are likely to need the packages from EPEL.)
Now back to LINBIT news. On Tuesday this week, we held the LINBIT Storage Day, partnering with Ampere and ShapeBlue to deliver a three-part webinar series on building next-generation IaaS. Thank you to everyone who attended and contributed to a fantastic event.
If you missed the LINBIT Storage Day, you will be happy to know you can still enjoy each webinar and experience the event from our website.
- I presented in the first webinar – ‘Updates To Apache CloudStack & LINBIT SDS’ – with ShapeBlue’s Giles Sirett.
- In the second webinar, ShapeBlue’s Andrija Panić and LINBIT’s Rene Peinthor discussed ‘CloudStack & LINBIT SDS Integration.’
- Finally, I took part in the ‘Ampere + LINBIT SDS + CloudStack’ panel discussion with ShapeBlue’s Ivet Petrova and Giles Sirett and Ampere’s Peter Pouliot and Christian Helmholz.
Virtual events can be challenging since you cannot see and feel the audience, as with in-person conferences. However, the number of quality questions we received after each session was really encouraging and gave a bit of a feel for the audience and how the sessions were received.
In other content-related news, Geo-clustering with Pacemaker & DRBD Proxy is a YouTube video that takes a deep dive into how multi-site Pacemaker clustering works and what makes it possible. Along with other learnings, we show that geo-clustering enables the pairing of High Availability with Disaster Recovery.
In addition, we have published three blog posts since my last newsletter. Compiling Your Own WinDRBD: 1.2 Branch Preview is a detailed guide covering everything from considerations before switching branches to new features and building WinDRBD on Linux.
Secondly, Open Source Innovations & Solutions at LINBIT is a fascinating post highlighting some lesser-known work the LINBIT developers have created. This post is an excellent example of what being open-source is all about. There may be a tool within the article that you would like to use – take a look!
Last but not least, Configuring Multipath iSCSI Targets in Pacemaker for Higher Availability demonstrates the upside of multipathing between an iSCSI target and initiator whenever possible. The combination of DRBD and Pacemaker leads to higher availability than just using one or the other.
In conclusion, I would like to share a few software updates the team has published since I last wrote.
drbd-reactor v1.3.0 may be of interest to both current drbd-reactor users as well as those looking for DRBD device monitoring via SNMP.
Linstor-server 1.25.0-rc.1 brings two critical improvements. First, from now on, it will handle failing backing storage devices by ignoring them after it sees an I/O failure once. Second, it makes sure that packets in LINSTOR’s TCP connections (control) have precedence over DRBD bulk (data) packets. That ensures LINSTOR stays connected with its satellites even when DRBD floods a network link with resync traffic. These two improvements are the results of unpleasant experiences in the field.
The python-linstor/linstor-client 1.20.0 release brings small features for the upcoming linstor-server 1.25.0 release.