SCiMMA

Expect service disruptions as we work to improve the stability of our service by improving the Kubernetes cluster and storage system.

On Tuesday, March 4, 2025 between 12pm and 1pm EST there will be a rolling restart of the dev and production kafka brokers to update certificates. After completion of this restart procedure, large message support will be enabled on Hopskotch. No interruption in service is planned. Users with long running Hop applications/connections that could be sensitive to a rolling restart may want to check them after the restarts.

Version 0.11.0 of the hop-client is now available on GitHub, from PyPI, and via Conda-Forge. Its most significant feature is is support for sending large messages using an auxilliary REST service which works alongside Kafka. Further changes and improvements are listed in the changelog.

On Tuesday, September 3, SCiMMA Hopskotch's mirroring of NASA's GCN) data streams will be upgraded to take its input from GCN's Kafka service.

The most major change will be that notices will no longer be a single stream, but a collection of distinct streams, as they are now presented by GCN over Kafka. The 'classic' data streams in both the Text and VOEvent formats will be mirrored (e.g. the gcn.classic.text.FERMI_GBM_ALERT and gcn.classic.voevent.FERMI_GBM_ALERT topics). Streams in the new JSON format will also be mirrored, e.g. gcn.notices.einstein_probe.wxt.alert. Republication of classic VOEvent messages on Hopskotch's gcn.notice topic will cease.

GCN Circulars will continue to be a single topic, however, they will move to the gcn.circulars topic in the new circular format). After the upgrade, no further messages will be published in the legacy format on the old gcn.circular topic.

The August 19 maintenance has concluded. As always, report any problems to support@scimma.org.

On August 19, we will conduct planned maintenance on SCiMMA infrastructure between 9 a.m. and 6 p.m. EDT. There may be short periods during this window when SCiMMA web services (https://my.hop.scimma.org and https://hermes.lco.global) will be briefly inaccessible or slow to respond. The total time when services are unavailable should be no more than a few minutes, and message traffic to and from the SCiMMA Kafka brokers should be unaffected.

On June 10, we will conduct planned maintenance on SCiMMA infrastructure between 9 a.m. and 6 p.m. EDT. There may be short periods during this window when SCiMMA web services (https://my.hop.scimma.org and https://hermes.lco.global) will be briefly inaccessible or slow to respond. The total time when services are unavailable should be no more than a few minutes, and message traffic to and from the SCiMMA Kafka brokers should be unaffected.

On March 5, we will conduct planned maintenance on SCiMMA infrastructure between 9 a.m. and 3 p.m. EST. There may be short periods during this window when valid publication or subscription requests to the Hopskotch Kafka service temporarily fail, and Hopskotch-connected web services (https://my.hop.scimma.org and https://hermes.lco.global) will be briefly inaccessible. The total time when services are unavailable should be no more than a few minutes, and some users of Kafka may not be affected.

On February 6, we will conduct planned maintenance on SCiMMA infrastructure between 9 a.m. and 3 p.m. EST. This will involve upgrades to systems which support identity management, so as a result there may be periods during this window when Keycloak (https://login.scimma.org/) is not available, and as a result web services using this identity provider (e.g. https://my.hop.scimma.org and https://hermes.lco.global) will not be accessible. There should be no impact on the SCiMMA Hopskotch service itself; its Kafka brokers should continue to handle messages as normal.