myasnikovdaniil · Nov 12, 2026 · coderabbitai · Nov 12, 2026
@@ -543,7 +543,7 @@ jobs:
     # Covers a green full suite. It does not cover a tenant-Kubernetes suite
     # running into its node-join deadline, so a red there can end as a
     # cancelled job. Held to one figure across every lane that runs the suite.
-    timeout-minutes: 215
+    timeout-minutes: 265
     permissions:
       contents: read
       packages: read

@@ -137,7 +137,7 @@ jobs:
     # Covers a green full suite. It does not cover a tenant-Kubernetes suite
     # running into its node-join deadline, so a red there can end as a
     # cancelled job. Held to one figure across every lane that runs the suite.
-    timeout-minutes: 215
+    timeout-minutes: 265
     permissions:
       contents: read
       checks: write # Breakpoint action updates its dedicated "Breakpoint Open" Check Run.
@@ -292,7 +292,7 @@ jobs:
       # collect-report packages cozyreport plus per-failed-test crust-gather
       # snapshots captured before Chainsaw cleanup.
       #
-      # A ceiling of its own, not the job's 215 minutes. `Upload cozyreport.tgz`
+      # A ceiling of its own, not the job's 265 minutes. `Upload cozyreport.tgz`
       # is the next step and the collector writes its tarball at the very end of
       # its run, so a read that never returns spends what is left of the job
       # budget, the job is cancelled, and the upload never runs -- the artifact

@@ -250,7 +250,7 @@ jobs:
     # Covers a green full suite. It does not cover a tenant-Kubernetes suite
     # running into its node-join deadline, so a red there can end as a
     # cancelled job. Held to one figure across every lane that runs the suite.
-    timeout-minutes: 215
+    timeout-minutes: 265
     permissions:
       contents: read
       packages: read
@@ -385,7 +385,7 @@ jobs:
           path: _out/chainsaw-report.xml
           if-no-files-found: ignore
 
-      # A ceiling of its own, not the job's 215 minutes. `Upload cozyreport.tgz`
+      # A ceiling of its own, not the job's 265 minutes. `Upload cozyreport.tgz`
       # is the next step and the collector writes its tarball at the very end of
       # its run, so a read that never returns spends what is left of the job
       # budget, the job is cancelled, and the upload never runs -- the artifact

@@ -953,7 +953,7 @@ jobs:
     # Covers a green full suite. It does not cover a tenant-Kubernetes suite
     # running into its node-join deadline, so a red there can end as a
     # cancelled job. Held to one figure across every lane that runs the suite.
-    timeout-minutes: 215
+    timeout-minutes: 265
     permissions:
       contents: read
       packages: read
@@ -1224,7 +1224,7 @@ jobs:
 
       # ▸ Collect debug information (always runs)
       #
-      #   A ceiling of its own, not the job's 215 minutes. `Upload
+      #   A ceiling of its own, not the job's 265 minutes. `Upload
       #   cozyreport.tgz` is the next step and the collector writes its tarball
       #   at the very end of its run, so a read that never returns spends what
       #   is left of the job budget, the job is cancelled, and the upload never
@@ -1235,7 +1235,7 @@ jobs:
       #   30 minutes is measured against the step: it takes 3-42s across sampled
       #   runs, on failed e2e jobs as well as green ones. It RESERVES nothing --
       #   `timeout-minutes` caps a step, it does not carve budget out of the job,
-      #   so a suite that has already burned 210 of the 215 leaves this step five
+      #   so a suite that has already burned 260 of the 265 leaves this step five
       #   minutes and the upload is cancelled with everything else. What the cap
       #   buys is that a hung read cannot be the reason.
       #   `continue-on-error` because the ceiling must not change what the job

@@ -121,7 +121,7 @@
 # the number; do not expect to find it written here.
 #
 # TWO ceilings sit outside it, and they fail differently. The job is capped at
-# `timeout-minutes: 215`, and reaching that cancels every step still to come --
+# `timeout-minutes: 265`, and reaching that cancels every step still to come --
 # the artifact upload among them. The step running this script is capped at 30 of
 # its own, and reaching THAT ends only this step: the uploads after it still run,
 # and `continue-on-error` on that step keeps the overrun out of the job's result,

@@ -208,7 +208,7 @@ _cozy_on_exit() {
       # capture would read as a complete one listing fewer crash-loops than
       # actually occurred. This handler's whole worst case is ~1350s (330 + 390
       # + 630, each leg including its 30s kill grace) against a job capped at
-      # 215 minutes with no per-step timeout —
+      # 265 minutes with no per-step timeout —
       # deliberate, for the same reason recorded in hack/e2e-chainsaw/.chainsaw.yaml:
       # a bounded, honest wait beats a kill that collects nothing.
       if [ "$_prev_rc" -ne 0 ]; then

@@ -66,14 +66,14 @@ spec:
   # script: it SIGKILLs the shell alone, orphaned children hold the pipe open,
   # and the op overruns (#4268 measured 606s under a 300s ceiling), so the
   # nominal ~5m was a figure a busy catch could exceed, not a cap. Tests run
-  # sequentially, and the E2E job is capped at 215 timeout-minutes. A cascade
+  # sequentially, and the E2E job is capped at 265 timeout-minutes. A cascade
   # reddening many suites can therefore spend a large share of the job in
   # diagnostics. That is deliberate — the old ceiling
   # bought its speed by collecting nothing at all — and it is bounded: both legs
   # only approach their ceilings on a genuinely degraded cluster, a systemic
   # failure usually reds the install gate and skips the E2E step entirely, and
   # the report-upload steps are `if: always()`. That last point does NOT extend
-  # to a job timeout: a 215-minute cap ends the job `cancelled`, and the ~4m48s
+  # to a job timeout: a 265-minute cap ends the job `cancelled`, and the ~4m48s
   # grace window is consumed by the earlier `Collect report` step, so both
   # cozyreport.tgz and the job log are lost (measured on runs 29557003429 and
   # 29568562337). If this ever does start truncating runs, bound the number of

@@ -35,7 +35,7 @@
 # and the teardown. This is the third suite in a full run that brings up a
 # tenant cluster, next to kubernetes-latest and kubernetes-previous. A failure
 # that stalls worker join everywhere reddens all three, and their red paths
-# together can exceed the E2E job's 215-minute cap, which cancels the job and
+# together can exceed the E2E job's 265-minute cap, which cancels the job and
 # loses its log and report. The cap is sized for the red path of the
 # kubernetes pair only; this suite is not counted in it, and that is an
 # accepted risk.

@@ -80,7 +80,8 @@ spec:
         # wait, all on the passing path. Raising the ceiling to cover them is not
         # available -- the guard in hack/run-kubernetes-serial-console_test.bats
         # holds COZY_OP_CEILING equal to BOTH suites' `timeout:`, and two suites
-        # at 79m plus their 22m teardowns exceed the job's own 215-minute cap.
+        # at 79m plus their 22m teardowns take 202m, which with the install and
+        # the rest of a full run does not fit the job's own 265-minute cap.
         # What is available is a measurement: on the green run of this branch
         # (33724179053) the whole Test took 569.72s against this 4020s op, with
         # kubernetes-previous, which carries no OIDC fold, at 470.19s. The