[
https://issues.apache.org/jira/browse/HADOOP-18675?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=17704537#comment-17704537
]
ASF GitHub Bot commented on HADOOP-18675:
-----------------------------------------
saxenapranav commented on code in PR #5508:
URL: https://github.com/apache/hadoop/pull/5508#discussion_r1147315190
##########
hadoop-tools/hadoop-azure/src/main/java/org/apache/hadoop/fs/azurebfs/utils/CachedSASToken.java:
##########
@@ -114,8 +125,20 @@ private static OffsetDateTime getExpiry(String token) {
OffsetDateTime seDate = OffsetDateTime.MIN;
try {
seDate = OffsetDateTime.parse(seValue, DateTimeFormatter.ISO_DATE_TIME);
- } catch (DateTimeParseException ex) {
- LOG.error("Error parsing se query parameter ({}) from SAS.", seValue,
ex);
+ } catch (DateTimeParseException dateTimeException) {
+ try {
+ TemporalAccessor dt = ISO_DATE_MIDNIGHT.parseBest(seValue,
OffsetDateTime::from, LocalDateTime::from);
+ if (dt instanceof OffsetDateTime) {
+ seDate = (OffsetDateTime) dt;
+ } else if (dt instanceof LocalDateTime) {
+ seDate = ((LocalDateTime) dt).atOffset(ZoneOffset.UTC);
+ } else {
+ throw dateTimeException;
+ }
+ } catch (DateTimeParseException dateOnlyException) {
Review Comment:
Can we have something like:
`private static final DateTimeFormatter[] formatters;`
```
private OffsetDateTime getParsedDateTime(String dateTime) {
for(DateTimeFormatter formatter : formatters) {
try {
TemporalAccessor temporalAccessor = formatter.parseBest(dateTime);
if(temporalAccessor instanceof OffsetDateTime) {
return (OffsetDateTime) temporalAccessor;
}
if(temporalAccessor instanceof LocalDateTime) {
return ((LocalDateTime) temporalAccessor).atOffset(ZoneOffset.UTC);
}
} catch (DateTimeParseException e) {
}
}
return null;
}
```
And from this method we will just call `seDate = getParsedDateTime(seValue);`
This way I feel we can prevent nested code.
> CachedSASToken noisy log errors when SAS token has YYYY-MM-DD expiration
> ------------------------------------------------------------------------
>
> Key: HADOOP-18675
> URL: https://issues.apache.org/jira/browse/HADOOP-18675
> Project: Hadoop Common
> Issue Type: Bug
> Components: fs/azure
> Reporter: Gil Cottle
> Priority: Major
> Labels: pull-request-available
>
> Error Description:
> When using SAS tokens with expiration dates in the format YYYY-MM-DD, a
> frequent error appears in the logs related to the date format. The error
> expects an ISO_DATE_TIME. See [existing
> implementation|https://github.com/apache/hadoop/blob/trunk/hadoop-tools/hadoop-azure/src/main/java/org/apache/hadoop/fs/azurebfs/utils/CachedSASToken.java#LL112-L119].
> The error is noisy in the logs, but does not cause issues.
> Example stacktrace:
> {code:java}
> 23/03/23 15:40:06 ERROR CachedSASToken: Error parsing se query parameter
> (2023-11-05) from SAS.
> java.time.format.DateTimeParseException: Text '2023-11-05' could not be
> parsed at index 10
> at
> java.time.format.DateTimeFormatter.parseResolved0(DateTimeFormatter.java:1949)
> at java.time.format.DateTimeFormatter.parse(DateTimeFormatter.java:1851)
> at java.time.OffsetDateTime.parse(OffsetDateTime.java:402)
> at
> shaded.databricks.azurebfs.org.apache.hadoop.fs.azurebfs.utils.CachedSASToken.getExpiry(CachedSASToken.java:116)
> at
> shaded.databricks.azurebfs.org.apache.hadoop.fs.azurebfs.utils.CachedSASToken.update(CachedSASToken.java:168)
> at
> shaded.databricks.azurebfs.org.apache.hadoop.fs.azurebfs.services.AbfsInputStream.readRemote(AbfsInputStream.java:670)
> at
> shaded.databricks.azurebfs.org.apache.hadoop.fs.azurebfs.services.ReadBufferWorker.lambda$run$0(ReadBufferWorker.java:66)
> at
> com.databricks.common.SparkTaskIOMetrics.withTaskIOMetrics(SparkTaskIOMetrics.scala:43)
> at
> shaded.databricks.azurebfs.org.apache.hadoop.fs.azurebfs.services.ReadBufferWorker.run(ReadBufferWorker.java:65)
> at java.lang.Thread.run(Thread.java:750) {code}
> Desired Resolution:
> Expiration code can read YYYY-MM-DD format as well as existing format.
--
This message was sent by Atlassian Jira
(v8.20.10#820010)
---------------------------------------------------------------------
To unsubscribe, e-mail: [email protected]
For additional commands, e-mail: [email protected]
