When srx chassis cluster failover not working is the actual symptom, redundancy-group 1 (and sometimes RG0) refuses to move to the other node even after a manual request chassis cluster failover or an interface going down that should have triggered it automatically. The most common causes, in order of how often we see them in the field, are interface monitoring weight exhaustion, a priority stuck at 0 from a prior failed monitor threshold, a down control or fabric link blocking the failover entirely, and a manual failover that was never cleared. This guide walks the check order before you start changing configuration on a live cluster.
This guide is written against Junos OS 21.4R3 on SRX300 and SRX1500 series branch and midrange platforms. Chassis cluster behavior on the SRX5000 line with multiple SPCs differs in some failover timing details, which is flagged where relevant.
What "Failover Not Working" Actually Looks Like
There are two different failure patterns that both get reported as "failover not working," and they have different root causes. The first is a manual failover request that silently does nothing -- the node stays primary. The second is an interface or link failure that should have triggered an automatic failover but did not, leaving traffic blackholed on the failed node. Separating which one you are dealing with is the first diagnostic step.
show chassis cluster status
This shows the current priority and redundancy-group ownership for both nodes. If RG1 priority reads 0 on the node you expect to be primary, you have found your answer immediately -- a priority of 0 means the node has deliberately removed itself from election and will not take over no matter what you request.
Diagnostic Order
show chassis cluster interfaces
Confirms the control link and fabric link states. If either shows down, Junos deliberately restricts failover behavior to avoid a split-brain scenario where both nodes believe they should be primary.
show chassis cluster statistics
Shows heartbeat and monitoring failure counts over time, useful for spotting a flapping interface monitor that has been silently dropping priority over days rather than all at once.
Root Cause 1: Interface Monitoring Weight Exhausted (Most Common)
SRX chassis cluster uses weighted interface monitoring on each redundancy group. Each monitored interface is assigned a weight, and when the sum of failed interface weights meets or exceeds the redundancy-group's configured threshold, that node's priority drops to 0 and it becomes ineligible to be primary. In our experience, this is the single most common reason a cluster appears to refuse failover -- the node is working correctly, it has simply disqualified itself because enough monitored interfaces are down.
show chassis cluster information redundancy-group 1
Compare the current weight total against the configured threshold. If a monitored interface went down for an unrelated reason (a misconfigured switchport, a bad SFP), that one failure can be enough to zero out the node's eligibility.
set chassis cluster redundancy-group 1 interface-monitor ge-0/0/1 weight 255
Verify the weight assignment matches your actual failover tolerance -- a single interface weighted at 255 with a threshold of 255 means that one link failure alone takes the node out of primary contention, which may or may not be what you intended.
Root Cause 2: Priority 0 From a Resolved Issue That Was Never Cleared
Juniper's own support documentation on troubleshooting Priority 0 issues on redundancy groups describes this exact scenario: the original interface failure is fixed, interfaces are physically up again, but the redundancy-group priority does not automatically recalculate back to its configured value. [Source: supportportal.juniper.net]
show interfaces terse | match ge-0/0/1
Confirm the previously failed interface is actually up now. If it is and priority is still 0, a manual reset of the redundancy-group monitoring state may be required rather than assuming it will self-heal.
Root Cause 3: Control Link or Fabric Link Down
The control link carries heartbeat and configuration synchronization traffic between the two nodes. The fabric link carries actual data-plane traffic between nodes for session state sync. If either is down, Junos restricts cluster behavior specifically to avoid both nodes claiming to be primary simultaneously, which can look like a stuck failover from the operator's perspective.
show chassis cluster interfaces | match "Control|Fabric"
A down control link is a physical and operational priority -- restore it before attempting any failover, manual or automatic. Running a cluster with a down control link, even briefly, risks a genuine split-brain condition.
Root Cause 4: A Previous Manual Failover Was Never Cleared
If someone ran request chassis cluster failover redundancy-group 1 node 1 during a maintenance window and never issued the corresponding reset, that manual override can persist and interfere with subsequent automatic failover behavior.
request chassis cluster failover reset redundancy-group 1
Always check for a stale manual override before assuming the cluster's automatic monitoring logic itself is broken.
Quick Reference: Symptom, Cause, Fix
| Symptom | Likely Cause | Fix |
|---|---|---|
| RG1 priority reads 0 on expected primary | Interface monitor weight exceeded threshold | Fix underlying interface, verify weight/threshold design |
| Interfaces back up but priority still 0 | Priority not auto-recalculated | Follow JTAC KB reset procedure for the redundancy group |
| Manual failover request does nothing | Control or fabric link down | Restore control/fabric link before attempting failover |
| Failover behaves oddly after a past maintenance window | Stale manual override | request chassis cluster failover reset |
Verification
show chassis cluster status
Cluster ID: 1
Node Priority Status Preempt Manual failover
redundancy-group: 0 , Failover count: 1
node0 200 primary no no
node1 100 secondary no no
redundancy-group: 1 , Failover count: 1
node0 200 primary yes no
node1 100 secondary yes no
A healthy cluster shows both nodes with nonzero priority, Manual failover reading no on both unless you intentionally triggered one, and the primary node matching the node with the higher configured priority. Always keep an out-of-band console connection open when testing failover on a production cluster -- a mistake here can take down both nodes' data plane simultaneously.
Full troubleshooting reference is in Juniper's TechLibrary. Troubleshoot SRX chassis cluster redundancy group not failing over
Frequently Asked Questions
Why does my SRX cluster not failover even when I run the manual command?
Check show chassis cluster status first for the priority on both nodes -- if the target node has priority 0, it is ineligible regardless of the manual request. Also confirm the control and fabric links are both up with show chassis cluster interfaces.
What does Priority 0 mean on an SRX redundancy group?
Priority 0 means the node has removed itself from primary election, almost always because the sum of failed interface-monitor weights met or exceeded the configured threshold. Fix the underlying interface failure and verify whether priority recalculates automatically or needs a manual reset.
How do I clear a manual failover on an SRX cluster?
Run request chassis cluster failover reset redundancy-group <group-number> to clear a previous manual override. Confirm the reset took effect with show chassis cluster status.
Does a down control link prevent failover?
Yes. Junos deliberately restricts cluster failover behavior when the control link is down to avoid a split-brain scenario where both nodes believe they are primary. Restore the control link before attempting any failover testing.
Why did OSPF reconvergence take so long after an SRX cluster failover?
This is a commonly reported issue separate from the failover mechanism itself -- OSPF with Fast Hello or BFD still has to rebuild adjacencies after the data plane moves to the new primary node. Tune BFD intervals carefully and confirm they are not set so aggressively that they themselves become a source of instability during the transition.
Related guides: see our security and firewalls category hub for more SRX-specific guides, our Junos VRRP not working guide for first-hop redundancy issues outside a chassis cluster, and our Junos LACP interface down guide if your cluster's fabric or control link itself is built on an aggregated bundle.
Juniper, Junos, MX, SRX, EX and QFX are trademarks of Juniper Networks, Inc. juniperclient.com is independent and unaffiliated.
Discussion (0)