summaryrefslogtreecommitdiff
path: root/issues/systems/reboot-tux01-tux02.gmi
diff options
context:
space:
mode:
authorArun Isaac2022-07-01 18:44:01 +0530
committerArun Isaac2022-07-01 18:44:01 +0530
commit2f84a722e5944bd4458abbd45fd637a9286bab04 (patch)
tree33e02614a9eb5e457c6162e8b2f67278ba07ddfe /issues/systems/reboot-tux01-tux02.gmi
parentc049be0d57f87151ad8db733150d9a49fb30ea31 (diff)
downloadgn-gemtext-2f84a722e5944bd4458abbd45fd637a9286bab04.tar.gz
Move issues lost inside the topics directory.
Diffstat (limited to 'issues/systems/reboot-tux01-tux02.gmi')
-rw-r--r--issues/systems/reboot-tux01-tux02.gmi54
1 files changed, 54 insertions, 0 deletions
diff --git a/issues/systems/reboot-tux01-tux02.gmi b/issues/systems/reboot-tux01-tux02.gmi
new file mode 100644
index 0000000..3186a0d
--- /dev/null
+++ b/issues/systems/reboot-tux01-tux02.gmi
@@ -0,0 +1,54 @@
+# Rebooting the GN production machine(s)
+
+I needed to add the hard disks in the BIOS to make them visible - one of the annoying aspects of these Dell machines. First on Tux02 I cheched the borg backups to see if we have a recent copy of MariaDB, GN2 etc. The DB is from 2 days ago and the genotypes of GN2 are a week old (because of a permission problem). I'll add a copy by hand of both - an opportunity to test the new 10Gbs router.
+
+Something funny is going on. When eno4 goes down the external webserver interface is not working. It appears, somehow, that 128.169.4.67 is covering for 128.169.5.59. I need to check that!
+
+# Tasks
+
+Before rebooting
+
+On Tux02:
+
+* [X] Check backups of DB and services
+* [X] Copy trees between machines
+
+On Tux01:
+
+* [ ] Network confused. See above.
+
+On both:
+
+* [X] Check network interface definitions (what happens on reboot)
+* [X] Check IPMI access - should get serial login
+
+
+# Info
+
+## Routing
+
+On tux02 eno2d1 is the 10Gbs network interface. Unfortunately I can't get it to connect at 10Gbs with Tux01 because the latter is using that port for the outside world.
+
+Playing with 10Gbs on Tux01 sent the hardware in a tail spin, what to think of this solution on
+
+```
+bnxt_en 0000:01:00.1 (unnamed net_device) (uninitialized): Error (timeout: 500015) msg {0x0 0x0} len:0
+
+Solution was to power down the server(s) and *remove* power cords for 5 minutes.
+```
+
+=> https://www.dell.com/community/PowerEdge-Hardware-General/Critical-network-bnxt-en-module-crashes-on-14G-servers/td-p/6031769
+
+The Linux kernel shows some fixes that are not on Tux01 yet
+
+=> https://lkml.org/lkml/2021/2/17/970
+
+In our case a simple reboot worked, fortunately.
+
+## Tags
+
+* assigned: pjotrp
+* status: unclear
+* priority: medium
+* type: system administration
+* keywords: systems, tux01, tux02, production