Skip to content

pxl_scripts/dx/shadow_trace: say that the protocol tabs need a trace - #118

Open
ConstanzeTU wants to merge 7 commits into
feat/pixie-native-sbobfrom
fix/shadow-trace-tabs-note
Open

pxl_scripts/dx/shadow_trace: say that the protocol tabs need a trace#118
ConstanzeTU wants to merge 7 commits into
feat/pixie-native-sbobfrom
fix/shadow-trace-tabs-note

Conversation

@ConstanzeTU

Copy link
Copy Markdown

Describes why the per-protocol tabs are empty until a trace is selected.

ConstanzeTU added 7 commits September 12, 2026 13:45
Signature verification is off on these clusters, so an unsigned authored
profile is adopted and governs. Counting signed == 1 reported zero for
containers that are governed. The panel now counts authored profiles and
reports verified and unsigned beside that count.
Shows every profile beside the authored profile governing the same
workload, with what the container did that the profile does not permit
and what it permits that was never used.
Names the two sides by what they mean and states that the authored side
changes through bobctl. learned_only is not ungoverned: every container
has a learned profile whether or not it is bound.
completion is set when the container starts; status only after the
sniffing window closes. A divergence against a profile still being
written is provisional, so the two columns are shown together and the
counts are split by whether the baseline has settled.
Asked over the picker's window, freshness returned an empty table whenever
the window held no rows, which is the reading it exists to rule out. It now
looks back over the retention period, so an empty result means the store
holds nothing rather than nothing recently.
…t#22

completion is set when a container starts and says nothing about whether
a profile governs. Records the duplicated-command defect as a known false
positive in the divergence panel rather than special-casing it.
The tabs match the session id exactly while the summary tables match on
substring, so at defaults the tabs are empty and the tables are not. The
dark surfaces also hold first-appearance content only, so an empty tab on
a busy trace is the design rather than a gap; the per-epoch tables answer
whether a surface is still collecting.
@ConstanzeTU
ConstanzeTU force-pushed the fix/shadow-trace-tabs-note branch from ac20538 to 5361392 Compare September 12, 2026 18:04
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Labels

None yet

Projects

None yet

Development

Successfully merging this pull request may close these issues.

1 participant