phet commented on code in PR #3610:
URL: https://github.com/apache/gobblin/pull/3610#discussion_r1041842571
##########
gobblin-metrics-libs/gobblin-metrics-base/src/main/avro/GaaSObservabilityEventV0.avsc:
##########
@@ -0,0 +1,162 @@
+{
+ "type": "record",
+ "name": "GaaSObservabilityEventV0",
+ "namespace" : "org.apache.gobblin.metrics",
+ "fields": [
+ {
+ "name": "timestamp",
+ "type": "long",
+ "doc": "Time at which event was created in millis"
+ }, {
+ "name" : "flowGroup",
+ "type" : "string",
+ "doc" : "Flow group for the GaaS flow",
+ "compliance" : "NONE"
+ }, {
+ "name" : "flowName",
+ "type" : "string",
+ "doc" : "Flow name for the GaaS flow",
+ "compliance" : "NONE"
+ }, {
+ "name" : "flowExecutionId",
+ "type" : "long",
+ "doc" : "Flow execution id for the GaaS flow",
+ "compliance" : "NONE"
+ }, {
+ "name": "jobSentToExecutor",
+ "type": "boolean",
+ "doc": "Whether or not this job was able to be sent to a job executor."
+ }, {
+ "name": "lastFlowModificationTime",
+ "type": "long",
+ "doc": "Timestamp in millis when the flow config was last modified"
+ }, {
+ "name" : "edgeId",
+ "type" : "string",
+ "doc" : "Flow edge id, in format <src_node>_<dest_node>_<edge_id>",
+ "compliance" : "NONE"
+ }, {
+ "name": "jobName",
+ "type": "string",
+ "doc": "The name of the Gobblin job, found in the job template. One edge
can contain multiple jobs",
+ "compliance" : "NONE"
+ }, {
+ "name": "jobType",
+ "type": {
+ "type": "enum",
+ "name": "JobType",
+ "symbols": [
+ "COPY",
+ "RETENTION",
+ "GOBBLIN"
+ ],
+ "symbolDocs": {
+ "COPY": "Gobblin distcp job",
+ "RETENTION": "Gobblin retention job",
+ "GOBBLIN": "Any Gobblin job"
+ }
+ },
+ "doc": "Gobblin job type running on GaaS, determined by the compiled job
template.",
+ "compliance": "NONE"
+ }, {
+ "name": "jobStatus",
+ "type": {
+ "type": "enum",
+ "name": "JobStatus",
+ "symbols": [
+ "SUCCEEDED",
+ "FAILED",
+ "CANCELLED"
+ ],
+ "doc": "Final job status for this job in the GaaS flow",
+ "compliance": "NONE"
+ }
+ }, {
+ "name": "jobStartTime",
+ "type": "long",
+ "doc": "Start time of the job in millis",
+ "compliance": "NONE"
+ }, {
+ "name": "jobEndTime",
+ "type": "long",
+ "doc": "Finish time of the job in millis",
+ "compliance": "NONE"
+ }, {
+ "name": "proxyUser",
+ "type": "string",
+ "doc": "Proxy user (if applicable) that ran the Gobblin job",
+ "compliance": "NONE"
+ }, {
+ "name": "executorLink",
+ "type": "string",
+ "doc": "Link to where the job is running, currently limited to Azkaban",
+ "compliance": "NONE"
+ }, {
+ "name": "executorId",
+ "type": "string",
Review Comment:
great Q on compilation failure! probably both ought to be null.
compilation error events should be purely for *scheduled flows that
previously compiled*--failure at submission time should be an interactive error
only, no event. nonetheless, given we need them, perhaps we're asking
`jobStatus.FAILURE` to do too much: it's already execution failure as well as
failure to send to executor--now compilation failure as well? it becomes a
status *category*, a catch-all. better might be another status, like `INVALID`
or else named `COMPILATION_FAILURE` / `UNRESOLVED` (?).
in fact, more I consider that, I'd seek also to discern `EXECUTION_FAILURE`
from `SUBMISSION_FAILURE` (actually I dislike the overloaded terminology, where
the user 'submits' a flow to gaas, and gaas 'submits' the job to the executor.)
other possible terms might be `QUEUEING_FAILURE` or `LOST` or `UNAVAILABLE`,
even `SKIPPED` if we consider possible quota overrun... but overall, whatever
the names, three distinct statuses for each kind of failure
I realize that would depart from borrowing a status previously defined in
the gaas job state .pdl, but faced w/ the alternative imprecision, I believe
that justifiable
--
This is an automated message from the Apache Git Service.
To respond to the message, please log on to GitHub and use the
URL above to go to the specific comment.
To unsubscribe, e-mail: dev-unsubscr...@gobblin.apache.org
For queries about this service, please contact Infrastructure at:
us...@infra.apache.org
