Bug 1204782
| Summary: | [RHEL 7.0 + 7.1] Host configure with DHCP is losing connectivity after some time - dhclient is not running | ||
|---|---|---|---|
| Product: | Red Hat Enterprise Virtualization Manager | Reporter: | rhev-integ |
| Component: | vdsm | Assignee: | Ido Barkan <ibarkan> |
| Status: | CLOSED ERRATA | QA Contact: | Michael Burman <mburman> |
| Severity: | urgent | Docs Contact: | |
| Priority: | urgent | ||
| Version: | 3.5.0 | CC: | adahms, bazulay, cshao, danken, ecohen, eedri, fdeutsch, gklein, hadong, huiwa, ibarkan, initscripts-maint-list, jentrena, jpopelka, kgoldbla, lbopf, lnykryn, lpeer, lsurette, lvernia, mburman, mkalinin, ovasik, pdwyer, sbonazzo, yaniwang, ycui, yeylon, ylavi |
| Target Milestone: | --- | Keywords: | ZStream |
| Target Release: | 3.5.1 | Flags: | ylavi:
Triaged+
|
| Hardware: | x86_64 | ||
| OS: | Linux | ||
| Whiteboard: | network | ||
| Fixed In Version: | vdsm-4.16.13-1.el6ev | Doc Type: | Bug Fix |
| Doc Text: |
After adopting Red Hat Enterprise Linux 7 and systemd, the work of preparing the environment before the VDSM service starts was managed by systemd itself using the ExecStartPre API. This preparation work was implemented under vdsm_init_common.sh and included network restoration. On DHCP networks, a side effect of this was the spawning of a dhclient daemon by init scripts. Since ExecStartPre API prohibits long-running processes, systemd kills those processes silently, leaving hosts with no dhclient to renew their IP lease. Now, the network restoration part of the preparation work is delegated to another dependent systemd unit resulting in a new 'oneshot' type service that is a vdsmd dependency, called 'vdsm-network'. Also systemd is now told not to kill dhclient processes upon stopping vdsm-network service.
|
Story Points: | --- |
| Clone Of: | 1187244 | Environment: | |
| Last Closed: | 2015-04-28 18:52:36 UTC | Type: | --- |
| Regression: | --- | Mount Type: | --- |
| Documentation: | --- | CRM: | |
| Verified Versions: | Category: | --- | |
| oVirt Team: | Network | RHEL 7.3 requirements from Atomic Host: | |
| Cloudforms Team: | --- | Target Upstream Version: | |
| Embargoed: | |||
| Bug Depends On: | 1187244 | ||
| Bug Blocks: | 1193058 | ||
|
Comment 1
Ido Barkan
2015-03-26 06:19:44 UTC
Verified and tested on both RHEL 7.0 + 7.1 rhevm-3.5.1-0.3.el6ev.noarch vdsm-4.16.13-1.el7ev.x86_64 dhclient-4.2.5-35.el7.x86_64 systemd-208-11.el7_0.6.x86_64 Ido, please supply doc text describes what's fixed here - you can use the bug this depends on as reference (it described the original issue and workarounds). After adopting rhel7 and systemd the work of preparing the environment before VDSM service start was managed by systemd itself using the ExecStartPre API. This preparation work was implemented under vdsm_init_common.sh and included network restoration. On DHCP networks, a side effect of this might be the spawning of a dhclient daemon by init scripts. Since ExecStartPre API prohibits long running processes as a result, systemd kills those processes silently, leaving hosts with no dhclient to renew their IP lease. The fix here was to delegate the network restoration part of the preparation work to another dependent systemd unit resulting in a new 'oneshot' type service that is a vdsmd dependency, called 'vdsm-network'. This was done in oVirt gerrit 29441. An additional change was telling systmd to not kill dhclient processes upon stopping vdsm-network service. This was implemented in oVirt gerrit 39213. Since the problem described in this bug report should be resolved in a recent advisory, it has been closed with a resolution of ERRATA. For information on the advisory, and where to find the updated files, follow the link below. If the solution does not work for you, open a new bug report. https://rhn.redhat.com/errata/RHBA-2015-0904.html |