Hristo Iliev
Hristo Iliev
Travis build job: https://travis-ci.org/cloudfoundry-incubator/cf-abacus/jobs/457640911 PR: https://github.com/cloudfoundry-incubator/cf-abacus/pull/1118 I can see at least two suspicious lines: ``` ... /bin/sh: 1: npm5: not found ... Error: Command failed: git commit -m "chore(package): update...
We need to provide a simplified deployment pipeline that demonstrates how we can deploy all Abacus components. Old pipeline (deleted with 3539bc1e27707c61d74b8263775f012d6793c307) was: * untested * using service instances we...
We can remove AUTH_SERVER env.variable and use only CF API. Having both might confuse integrators (#739).
Now cf-renewer reads all started apps from the last month and tries to send the refreshed usage to the current month. This loads Abacus unnecessarily. One solution would be to...
Batch endpoint does not enforce the max batch size and we can have different batch sizes from different clients. The endpoint in router module does not use BATCH_SIZE. Still it...
vcapenv removes the schema and domain from proper host names For example `https://.` will become `https://.` after processing. This forces users to specify hostnames with ports like `https://.:443`
If a resource provider needs to define its own schema for one of the fields in the usage doc (account, org, space, consumer) it can do so with URL-like string....
#601 proposes /health endpoint. Usually we wrap endpoints with breaker to use performance counters and feed it with data to open in case of a problem. To have long-running apps...
Setting `SLACK=2M` now equals `SLACK=62d`. This is mostly true except for February. We should calculate the slack with moment as it gives accurate results for slack windows spanning across February.
When running dataflow with DB_PARTITIONS > 1 replay cannot find all documents. dataflow supports a single input db assigned to app instance and therefore we use hardcoded number of partitions...