using pxe boot, the etcd always "heartbeat near election timeout" #963
Comments
|
any suggestion ？ |
|
May you check the machine list? |
|
check from both machines: core@localhost ~ $ curl http://127.0.0.1:7001/v2/admin/machines I can set and get on leader, only can get on follwer. |
|
yesterday, I use -name whit IP address, like -name 192.168.1.4. I change the it as -name 4, Now the but when I set some message on follower, still get error. |
|
and log on follower. core@localhost ~ $ journalctl -fu config-etcd.service log on leader, core@localhost ~ $ journalctl -fu config-etcd.service |
|
etcdctl uses clientURL to connect to each machine. |
|
it works correctly, thanks!!! |
I setup PXE server in a subnet, use it to boot coreOS. Since "$public_ipv4" dosen't support in PXE,
so I use could-config to start up etcd. My cloud-config file is as below:
cloud-config
coreos:
units:
- name: etcd.service
command: stop
- name: config-etcd.service
command: start
content: |
[Unit]
Description=Config Etcd
After=etcd.service
ssh_authorized_keys:
in could-config I download a script, use it to get IP address and start etcd, the script is as below.
#!/bin/bash
export publicIP=
ifconfig eno1 | sed -n 2p | awk '{ print $2 }'systemctl stop etcd
etcd -name $publicIP -peer-addr $publicIP:7001 -addr 127.0.0.1:4001 -discovery http://192.168.1.2:4001/v2/keys/0000040 -peer- election-timeout=5000 -peer-heartbeat-interval=5000 snapshot=true -v
after the coreOS is boot up, the script is downloaded, and it work as expecting. But the etcd always
get timeout. I have two nodes, the follower always get timeout, I mean no matter which is become
follower, and the leader is work fine.
core@localhost ~ $ journalctl -fu config-etcd.service
-- Logs begin at Wed 2014-08-27 08:19:35 UTC. --
Aug 27 08:20:02 localhost startEtcd.sh[597]: [etcd] Aug 27 08:20:02.753 INFO | 192.168.1.5: warning: heartbeat near election timeout: 4.999946418s
Aug 27 08:20:07 localhost startEtcd.sh[597]: [etcd] Aug 27 08:20:07.753 INFO | 192.168.1.5: warning: heartbeat near election timeout: 4.999720814s
Aug 27 08:20:12 localhost startEtcd.sh[597]: [etcd] Aug 27 08:20:12.753 INFO | 192.168.1.5: warning: heartbeat near election timeout: 4.999857421s
Aug 27 08:20:17 localhost startEtcd.sh[597]: [etcd] Aug 27 08:20:17.753 INFO | 192.168.1.5: warning: heartbeat near election timeout: 4.999935035s
Aug 27 08:20:22 localhost startEtcd.sh[597]: [etcd] Aug 27 08:20:22.753 INFO | 192.168.1.5: warning: heartbeat near election timeout: 5.000014421s
Aug 27 08:20:27 localhost startEtcd.sh[597]: [etcd] Aug 27 08:20:27.753 INFO | 192.168.1.5: warning: heartbeat near election timeout: 4.999861909s
Aug 27 08:20:32 localhost startEtcd.sh[597]: [etcd] Aug 27 08:20:32.753 INFO | 192.168.1.5: warning: heartbeat near election timeout: 4.999855009s
Aug 27 08:20:37 localhost startEtcd.sh[597]: [etcd] Aug 27 08:20:37.753 INFO | 192.168.1.5: warning: heartbeat near election timeout: 4.999945439s
Aug 27 08:20:42 localhost startEtcd.sh[597]: [etcd] Aug 27 08:20:42.753 INFO | 192.168.1.5: warning: heartbeat near election timeout: 4.999893319s
Aug 27 08:20:47 localhost startEtcd.sh[597]: [etcd] Aug 27 08:20:47.753 INFO | 192.168.1.5: warning: heartbeat near election timeout: 4.99987582s
Aug 27 08:20:52 localhost startEtcd.sh[597]: [etcd] Aug 27 08:20:52.753 INFO | 192.168.1.5: warning: heartbeat near election timeout: 4.999934057s
if I set something on the follower node, will occur error, see below,
core@localhost ~ $ etcdctl set /message test
Error: 501: All the given peers are not reachable (Tried to connect to each peer twice and failed) [0]
I set something on the leader, everything is working fine, the follower can also get the the message
using etcdctl.
I tried all version list in page https://coreos.com/docs/running-coreos/bare-metal/booting-with-pxe/,
stable, beta, alpha, all have the same issue. But no matter what version I use to boot, when login
it always show CoreOS (beta).
I also tuned the parameters with etcd "election-timeout=5000 -peer-heartbeat-interval=5000", but
the issue still there.
googled, there are some similar issue like #868, #594, #915, I am not sure if these bug is the same
with mine.
I also run 5 coreOS nodes cluser with vagrant and virtualbox, everything works fine. But when longin
it show CoreOS(alpha).
Did I do anything wrong ? And how I make it work ?
The text was updated successfully, but these errors were encountered: