Bug #3431
MultiWAN provider remains disabled after failure
Status: | CLOSED | Start date: | ||
---|---|---|---|---|
Priority: | Normal | Due date: | ||
Assignee: | - | % Done: | 100% | |
Category: | nethserver-lsm | |||
Target version: | v6.8 | |||
Security class: | Resolution: | |||
Affected version: | NEEDINFO: | No |
Description
Sometimes, after a link failure, a provider in a multiwan setup doesn't come up after the problem is resolved.
I'm able to reproduce this issue only using a specific adsl router (leased by a big Italian isp).
Steps to reproduce:
1. in a multiwan setup simulate a failure disconnecting the cable from red to adsl modem
2. apply some changes to the firewall, I use restart lsm
3. wait and see: the provider never comes back (I use shorewall status -i)
I found that LSM (Link Status Monitor) uses arp packets instead of icmp to discover link status in this scenario.
I also found that configuring LSM to use the sourceip address of the network interface forces icmp packets, solving the issue.
Associated revisions
lsm.conf: use sourceip when avaialable. Refs #3431
lsm-wan-notify: do not notify link down on lsm restart. Refs #3431
lsm-wan-notify: do not notify link down on lsm restart. Refs #3431
lsm.conf: use sourceip when avaialable. Refs #3431
lsm.conf: use empty sourceip for dynamic links (PPPoE). Refs #3431
lsm.conf: use empty sourceip for dynamic links (PPPoE). Refs #3431
History
#1 Updated by Filippo Carletti over 4 years ago
Note: while testing, I found that lsm notifies link down on every restart (while it doesn't, in case of link up due to the unknown_up_notify option).
#2 Updated by Giacomo Sanchietti over 4 years ago
- Status changed from NEW to TRIAGED
- % Done changed from 0 to 20
#3 Updated by Giacomo Sanchietti over 4 years ago
- Status changed from TRIAGED to ON_DEV
- % Done changed from 20 to 30
#4 Updated by Filippo Carletti over 4 years ago
- Status changed from ON_DEV to MODIFIED
- % Done changed from 30 to 60
#5 Updated by Filippo Carletti over 4 years ago
- Status changed from MODIFIED to ON_QA
- % Done changed from 60 to 70
In nethserver-testing:
nethserver-lsm-1.1.2-1.2.gaf7fbc4.ns6.noarch.rpm
Test case:
1. update
2. simulate link failure
3. restart lsm
4. re-establish link
Expected behaviour:
1. if mail notification is enabled, you should receive one down mail on link failure and one up mail on link connection
2. you shouldn't receive email when you restart lsm
3. shorewall status -i should report real status
All tests should be repeated on a system where one of the uplink uses a dhcp dynamic address and one where pppoe is used.
#6 Updated by Filippo Carletti over 4 years ago
- Status changed from ON_QA to TRIAGED
- % Done changed from 70 to 20
PPPoE is lo longer monitored after update.
#7 Updated by Filippo Carletti over 4 years ago
- Status changed from TRIAGED to MODIFIED
- % Done changed from 20 to 60
#8 Updated by Filippo Carletti over 4 years ago
- Status changed from MODIFIED to ON_QA
- % Done changed from 60 to 70
Updated package:
nethserver-lsm-1.1.2-1.3.g8974c30.ns6.noarch.rpm
#9 Updated by Davide Marini over 4 years ago
- Status changed from ON_QA to VERIFIED
- % Done changed from 70 to 90
Tested in many different MultiWan configurations, the new rpm always solved the problem.
#10 Updated by Giacomo Sanchietti over 4 years ago
- Status changed from VERIFIED to CLOSED
- % Done changed from 90 to 100
Released:
nethserver-lsm-1.1.3-1.ns6.noarch.rpm