[jira] [Updated] (HIVE-17114) HoS: Possible skew in shuffling when data is not really skewed
[ https://issues.apache.org/jira/browse/HIVE-17114?page=com.atlassian.jira.plugin.system.issuetabpanels:all-tabpanel ] Rui Li updated HIVE-17114: -- Resolution: Fixed Fix Version/s: 3.0.0 Status: Resolved (was: Patch Available) Pushed to master. Thanks Chao for the review. > HoS: Possible skew in shuffling when data is not really skewed > -- > > Key: HIVE-17114 > URL: https://issues.apache.org/jira/browse/HIVE-17114 > Project: Hive > Issue Type: Bug >Reporter: Rui Li >Assignee: Rui Li >Priority: Minor > Fix For: 3.0.0 > > Attachments: HIVE-17114.1.patch, HIVE-17114.2.patch, > HIVE-17114.3.patch > > > Observed in HoS and may apply to other engines as well. > When we join 2 tables on a single int key, we use the key itself as hash code > in {{ObjectInspectorUtils.hashCode}}: > {code} > case INT: > return ((IntObjectInspector) poi).get(o); > {code} > Suppose the keys are different but are all some multiples of 10. And if we > choose 10 as #reducers, the shuffle will be skewed. -- This message was sent by Atlassian JIRA (v6.4.14#64029)
[jira] [Updated] (HIVE-17114) HoS: Possible skew in shuffling when data is not really skewed
[ https://issues.apache.org/jira/browse/HIVE-17114?page=com.atlassian.jira.plugin.system.issuetabpanels:all-tabpanel ] Rui Li updated HIVE-17114: -- Attachment: HIVE-17114.3.patch > HoS: Possible skew in shuffling when data is not really skewed > -- > > Key: HIVE-17114 > URL: https://issues.apache.org/jira/browse/HIVE-17114 > Project: Hive > Issue Type: Bug >Reporter: Rui Li >Assignee: Rui Li >Priority: Minor > Attachments: HIVE-17114.1.patch, HIVE-17114.2.patch, > HIVE-17114.3.patch > > > Observed in HoS and may apply to other engines as well. > When we join 2 tables on a single int key, we use the key itself as hash code > in {{ObjectInspectorUtils.hashCode}}: > {code} > case INT: > return ((IntObjectInspector) poi).get(o); > {code} > Suppose the keys are different but are all some multiples of 10. And if we > choose 10 as #reducers, the shuffle will be skewed. -- This message was sent by Atlassian JIRA (v6.4.14#64029)
[jira] [Updated] (HIVE-17114) HoS: Possible skew in shuffling when data is not really skewed
[ https://issues.apache.org/jira/browse/HIVE-17114?page=com.atlassian.jira.plugin.system.issuetabpanels:all-tabpanel ] Rui Li updated HIVE-17114: -- Attachment: HIVE-17114.2.patch Patch v2 updates some golden files. Most of the changes are because we'll use different vector RS operators with uniform hash, which is expected. > HoS: Possible skew in shuffling when data is not really skewed > -- > > Key: HIVE-17114 > URL: https://issues.apache.org/jira/browse/HIVE-17114 > Project: Hive > Issue Type: Bug >Reporter: Rui Li >Assignee: Rui Li >Priority: Minor > Attachments: HIVE-17114.1.patch, HIVE-17114.2.patch > > > Observed in HoS and may apply to other engines as well. > When we join 2 tables on a single int key, we use the key itself as hash code > in {{ObjectInspectorUtils.hashCode}}: > {code} > case INT: > return ((IntObjectInspector) poi).get(o); > {code} > Suppose the keys are different but are all some multiples of 10. And if we > choose 10 as #reducers, the shuffle will be skewed. -- This message was sent by Atlassian JIRA (v6.4.14#64029)
[jira] [Updated] (HIVE-17114) HoS: Possible skew in shuffling when data is not really skewed
[ https://issues.apache.org/jira/browse/HIVE-17114?page=com.atlassian.jira.plugin.system.issuetabpanels:all-tabpanel ] Rui Li updated HIVE-17114: -- Status: Patch Available (was: Open) > HoS: Possible skew in shuffling when data is not really skewed > -- > > Key: HIVE-17114 > URL: https://issues.apache.org/jira/browse/HIVE-17114 > Project: Hive > Issue Type: Bug >Reporter: Rui Li >Assignee: Rui Li >Priority: Minor > Attachments: HIVE-17114.1.patch > > > Observed in HoS and may apply to other engines as well. > When we join 2 tables on a single int key, we use the key itself as hash code > in {{ObjectInspectorUtils.hashCode}}: > {code} > case INT: > return ((IntObjectInspector) poi).get(o); > {code} > Suppose the keys are different but are all some multiples of 10. And if we > choose 10 as #reducers, the shuffle will be skewed. -- This message was sent by Atlassian JIRA (v6.4.14#64029)
[jira] [Updated] (HIVE-17114) HoS: Possible skew in shuffling when data is not really skewed
[ https://issues.apache.org/jira/browse/HIVE-17114?page=com.atlassian.jira.plugin.system.issuetabpanels:all-tabpanel ] Rui Li updated HIVE-17114: -- Attachment: HIVE-17114.1.patch I think we can set the UNIFORM trait to the RS and then MurmurHash is used to compute the hash code, which can solve the issue here. > HoS: Possible skew in shuffling when data is not really skewed > -- > > Key: HIVE-17114 > URL: https://issues.apache.org/jira/browse/HIVE-17114 > Project: Hive > Issue Type: Bug >Reporter: Rui Li >Assignee: Rui Li >Priority: Minor > Attachments: HIVE-17114.1.patch > > > Observed in HoS and may apply to other engines as well. > When we join 2 tables on a single int key, we use the key itself as hash code > in {{ObjectInspectorUtils.hashCode}}: > {code} > case INT: > return ((IntObjectInspector) poi).get(o); > {code} > Suppose the keys are different but are all some multiples of 10. And if we > choose 10 as #reducers, the shuffle will be skewed. -- This message was sent by Atlassian JIRA (v6.4.14#64029)