XIO

xWire

Posts from every feed QUEST has found across the web — blogs, news, podcasts, channels and more.

ISC Open Source Projects activity

Razvan Becheriu closed issue #4456: Kea can log garbage from clients when logging packet contents at ISC Open Source Projects / Kea

Describe the bug There's two related issues here: Kea always directly logs options defined as strings, even if they contain non-printable characters Kea logs OpaqueDataTuple as a string, despite being called "opaque" In the customer's case, the first was observed with option 12 (hostname), and the second was observed with option 124 (vendor-independent vendor class). To Reproduce Steps to reproduce the behavior: Run a DHCP daemon with debug logging turned up high enough to log packet contents…

Francis Dupont commented on merge request !3337 at ISC Open Source Projects / Kea

Argh, got a new failure: [ RUN ] D2StatsTest.names ../src/lib/d2srv/tests/d2_tsig_key_unittest.cc:29: Failure Expected equality of these values: 4U Which is: 4 D2Stats::ncr.size() Which is: 5 [ FAILED ] D2StatsTest.names (0 ms) I think it is trivial to fix as this ticket added a new stat but it requires another round of review...

Razvan Becheriu closed issue #4736: RADIUS hook: access backend never reactivates after any dynamic reconfiguration (config-reload/config-set) at ISC Open So...

RADIUS hook: access backend never reactivates after any dynamic reconfiguration (config-reload/config-set) Component: libdhcp_radius.so (src/hooks/dhcp/radius) Kea version: 3.2.0 (also expected to affect other 3.x releases - the affected code is unchanged on current master as of this report) OS/build: Debian-based Docker image (jonasal/kea-dhcp4:3, resolves to the same digest as the 3.2.0 tag - confirmed non-Alpine, glibc, not musl) Severity: High - silent, total loss of RADIUS-based client…

Aakash opened issue #6509: DELEG/DELEGPARAM server-name text form is not round-trip safe: decimal escapes and embedded dots are silently corrupted (survives ...

Summary DELEG/DELEGPARAM server-name= values are not safe to serialize to text and parse back. A byte written as a decimal escape is emitted double-escaped and then re-parsed as literal digits (\140 -> 140), and a dot inside a label is emitted in a form that re-parses as a label separator (a.b as one label -> a,b). Round-tripping silently alters the RDATA, so every text dump/reload cycle (zone dump with masterfile-format text, named-compilezone -f text, rndc dumpdb, dig output) produces…

David Peterson commented on issue #2807 at ISC Open Source Projects / stork

Follow-up: fork attempt failed with: Limit reached You cannot create projects in your personal namespace. Contact your GitLab administrator. @tomek @slawek @marcin @vicky - please grant a project allocation / permission to fork isc-projects/stork into my namespace (theorem) so I can push the prepared branch and open an MR. The change is already cherry-picked onto current master locally (commit subject: Add PostgreSQL lease listing and read-only forensic search).

David Peterson opened issue #2807: PostgreSQL lease listing and read-only forensic (legal-log) search at ISC Open Source Projects / stork

Summary Stork currently only pulls leases from Kea memfile backends. Many deployments (including ours) store leases and forensic/legal-log history in PostgreSQL. This issue proposes: Teaching the agent/leases puller to read Kea PostgreSQL lease4/lease6 using the same DB credentials Kea uses to write, via a forced read-only session. Adding a date-bounded Forensic Search (UI + REST + gRPC) for Kea legal-log PostgreSQL history. Start and end times are required so Stork cannot scan the entire…

BJ Hinkle commented on issue #4896 at ISC Open Source Projects / Kea

Hi Darren, Thanks for looking into it. Certainly! I've attached an archive that has everything from one failing run: the later run mentioned in the description, where we read status-get and tried ha-scopes (2026-11-05, 16:20 to 16:38 UTC, 20 takeovers, 2 of them bad). kea-31-kea-dhcp4.conf and kea-32-kea-dhcp4.conf: each server's configuration as loaded for the run. Both pass kea-dhcp4 -t. The subnets normally come from an include file, which is inlined here. Left out of it: three site subnets…

Thorsten Glaser commented on issue #4893 at ISC Open Source Projects / Kea

Is it possible that this also occurs without dynamic reconfiguration, but in an HA setup? In https://codeberg.org/mirabilos/kea-dhcp/issues/1#issuecomment-27320385, Václav wrote: @mirabilos ( https://codeberg.org/mirabilos ) wrote in #1 (closed) (komentář) ( https://codeberg.org/mirabilos/kea-dhcp/issues/1#issuecomment-27218109 ) : This appears to be caused by a bug in KEA: #4893 Do you use reconfiguration? I don't know what exactly you mean. But now I did: stop standby server stop primary…

Piotrek Zadroga commented on merge request !1526 at ISC Open Source Projects / stork

Thanks for suggestion but I think that "local" in the original description referred to the local time zone, not the locale. The moment-based version converted to the browser's time zone (moment.tz(guess)) and always used the fixed YYYY-MM-DD HH:mm:ss[.SSS] format. This implementation keeps that output exactly, so nothing changes for users. return date.toLocaleString() => that would change this function behavior and I'd like to avoid that. return ${isoDate.substr(0,10)} ${isoDate.substr(11,12)}…