[jira] [Created] (FLINK-35876) mysql cdc may change from snapshot mode to binlog mode failed when cpu load of mysql server is 100%
Youle created FLINK-35876: - Summary: mysql cdc may change from snapshot mode to binlog mode failed when cpu load of mysql server is 100% Key: FLINK-35876 URL: https://issues.apache.org/jira/browse/FLINK-35876 Project: Flink Issue Type: Bug Reporter: Youle -- This message was sent by Atlassian Jira (v8.20.10#820010)
[jira] [Updated] (FLINK-35876) mysql cdc may change from snapshot mode to binlog mode failed when cpu load of mysql server is 100%
[ https://issues.apache.org/jira/browse/FLINK-35876?page=com.atlassian.jira.plugin.system.issuetabpanels:all-tabpanel ] Youle updated FLINK-35876: -- Attachment: image-2024-07-23-15-47-10-376.png Description: I use mysql cdc to read data from mysql to write to doris, mysql cdc may change from sanpshot mode to binlog mode failed when I set up 2 flink job(20 parallism of source) and the cpu load of mysql server is 100%. It can't change even the cpu load drop and just hang up, won't consume latest binlog data. But when I only set up one flink job(20 parallism of source) and the cpu load of mysql server is less than 100%, that mistake never happens. Is there anything to do to avoid this mistake if we can't promise the cpu load always in a healthy state? !image-2024-07-23-15-47-10-376.png! > mysql cdc may change from snapshot mode to binlog mode failed when cpu load > of mysql server is 100% > --- > > Key: FLINK-35876 > URL: https://issues.apache.org/jira/browse/FLINK-35876 > Project: Flink > Issue Type: Bug >Reporter: Youle >Priority: Major > Attachments: image-2024-07-23-15-47-10-376.png > > > I use mysql cdc to read data from mysql to write to doris, mysql cdc may > change from sanpshot mode to binlog mode failed when I set up 2 flink job(20 > parallism of source) and the cpu load of mysql server is 100%. It can't > change even the cpu load drop and just hang up, won't consume latest binlog > data. But when I only set up one flink job(20 parallism of source) and the > cpu load of mysql server is less than 100%, that mistake never happens. > Is there anything to do to avoid this mistake if we can't promise the cpu > load always in a healthy state? > !image-2024-07-23-15-47-10-376.png! -- This message was sent by Atlassian Jira (v8.20.10#820010)
[jira] [Updated] (FLINK-35876) mysql cdc may change from snapshot mode to binlog mode failed when cpu load of mysql server is 100%
[ https://issues.apache.org/jira/browse/FLINK-35876?page=com.atlassian.jira.plugin.system.issuetabpanels:all-tabpanel ] Youle updated FLINK-35876: -- Description: I use mysql cdc to read data from mysql to write to doris, mysql cdc may change from sanpshot mode to binlog mode failed when I set up 2 flink job(20 parallism of source) and the cpu load of mysql server is 100%. It can't change even the cpu load drop and just hang up, won't consume latest binlog data and there is no exception information in logs. But when I only set up one flink job(20 parallism of source) and the cpu load of mysql server is less than 100%, that mistake never happens. Is there anything to do to avoid this mistake if we can't promise the cpu load always in a healthy state? !image-2024-07-23-15-47-10-376.png! was: I use mysql cdc to read data from mysql to write to doris, mysql cdc may change from sanpshot mode to binlog mode failed when I set up 2 flink job(20 parallism of source) and the cpu load of mysql server is 100%. It can't change even the cpu load drop and just hang up, won't consume latest binlog data. But when I only set up one flink job(20 parallism of source) and the cpu load of mysql server is less than 100%, that mistake never happens. Is there anything to do to avoid this mistake if we can't promise the cpu load always in a healthy state? !image-2024-07-23-15-47-10-376.png! > mysql cdc may change from snapshot mode to binlog mode failed when cpu load > of mysql server is 100% > --- > > Key: FLINK-35876 > URL: https://issues.apache.org/jira/browse/FLINK-35876 > Project: Flink > Issue Type: Bug >Reporter: Youle >Priority: Major > Attachments: image-2024-07-23-15-47-10-376.png > > > I use mysql cdc to read data from mysql to write to doris, mysql cdc may > change from sanpshot mode to binlog mode failed when I set up 2 flink job(20 > parallism of source) and the cpu load of mysql server is 100%. It can't > change even the cpu load drop and just hang up, won't consume latest binlog > data and there is no exception information in logs. But when I only set up > one flink job(20 parallism of source) and the cpu load of mysql server is > less than 100%, that mistake never happens. > Is there anything to do to avoid this mistake if we can't promise the cpu > load always in a healthy state? > !image-2024-07-23-15-47-10-376.png! -- This message was sent by Atlassian Jira (v8.20.10#820010)
[jira] [Updated] (FLINK-35876) mysql cdc may change from snapshot mode to binlog mode failed when cpu load of mysql server is 100%
[ https://issues.apache.org/jira/browse/FLINK-35876?page=com.atlassian.jira.plugin.system.issuetabpanels:all-tabpanel ] Youle updated FLINK-35876: -- Description: I use mysql cdc to read data from mysql to write to doris, mysql cdc may change from sanpshot mode to binlog mode failed when I set up 2 flink job(20 parallism of source) and the cpu load of mysql server is 100%. It can't change even the cpu load drop and just hang up, won't consume latest binlog data and there is no exception information in logs. But when I only set up one flink job(20 parallism of source) and the cpu load of mysql server is less than 100%, that mistake never happens. mysql table: about 250 million data flink version: 1.18.1 mysql cdc connector version: 3.1.0 doris connector version :1.18-1.6.2 Is there anything to do to avoid this mistake if we can't promise the cpu load always in a healthy state? !image-2024-07-23-15-47-10-376.png! was: I use mysql cdc to read data from mysql to write to doris, mysql cdc may change from sanpshot mode to binlog mode failed when I set up 2 flink job(20 parallism of source) and the cpu load of mysql server is 100%. It can't change even the cpu load drop and just hang up, won't consume latest binlog data and there is no exception information in logs. But when I only set up one flink job(20 parallism of source) and the cpu load of mysql server is less than 100%, that mistake never happens. Is there anything to do to avoid this mistake if we can't promise the cpu load always in a healthy state? !image-2024-07-23-15-47-10-376.png! > mysql cdc may change from snapshot mode to binlog mode failed when cpu load > of mysql server is 100% > --- > > Key: FLINK-35876 > URL: https://issues.apache.org/jira/browse/FLINK-35876 > Project: Flink > Issue Type: Bug >Reporter: Youle >Priority: Major > Attachments: image-2024-07-23-15-47-10-376.png > > > I use mysql cdc to read data from mysql to write to doris, mysql cdc may > change from sanpshot mode to binlog mode failed when I set up 2 flink job(20 > parallism of source) and the cpu load of mysql server is 100%. It can't > change even the cpu load drop and just hang up, won't consume latest binlog > data and there is no exception information in logs. But when I only set up > one flink job(20 parallism of source) and the cpu load of mysql server is > less than 100%, that mistake never happens. > mysql table: about 250 million data > flink version: 1.18.1 > mysql cdc connector version: 3.1.0 > doris connector version :1.18-1.6.2 > > Is there anything to do to avoid this mistake if we can't promise the cpu > load always in a healthy state? > !image-2024-07-23-15-47-10-376.png! -- This message was sent by Atlassian Jira (v8.20.10#820010)
[jira] [Updated] (FLINK-35876) mysql cdc may change from snapshot mode to binlog mode failed when cpu load of mysql server is 100%
[ https://issues.apache.org/jira/browse/FLINK-35876?page=com.atlassian.jira.plugin.system.issuetabpanels:all-tabpanel ] Youle updated FLINK-35876: -- Component/s: Flink CDC Affects Version/s: cdc-3.1.0 > mysql cdc may change from snapshot mode to binlog mode failed when cpu load > of mysql server is 100% > --- > > Key: FLINK-35876 > URL: https://issues.apache.org/jira/browse/FLINK-35876 > Project: Flink > Issue Type: Bug > Components: Flink CDC >Affects Versions: cdc-3.1.0 >Reporter: Youle >Priority: Major > Attachments: image-2024-07-23-15-47-10-376.png > > > I use mysql cdc to read data from mysql to write to doris, mysql cdc may > change from sanpshot mode to binlog mode failed when I set up 2 flink job(20 > parallism of source) and the cpu load of mysql server is 100%. It can't > change even the cpu load drop and just hang up, won't consume latest binlog > data and there is no exception information in logs. But when I only set up > one flink job(20 parallism of source) and the cpu load of mysql server is > less than 100%, that mistake never happens. > mysql table: about 250 million data > flink version: 1.18.1 > mysql cdc connector version: 3.1.0 > doris connector version :1.18-1.6.2 > > Is there anything to do to avoid this mistake if we can't promise the cpu > load always in a healthy state? > !image-2024-07-23-15-47-10-376.png! -- This message was sent by Atlassian Jira (v8.20.10#820010)
[jira] [Updated] (FLINK-35876) mysql cdc may change from snapshot mode to binlog mode failed when cpu load of mysql server is 100%
[ https://issues.apache.org/jira/browse/FLINK-35876?page=com.atlassian.jira.plugin.system.issuetabpanels:all-tabpanel ] Youle updated FLINK-35876: -- Description: I use mysql cdc to read data from mysql to write to doris, mysql cdc may change from sanpshot mode to binlog mode failed when I set up 2 flink job(20 parallelism of source) and the cpu load of mysql server is 100%. It can't change even the cpu load drop and just hang up, won't consume latest binlog data and there is no exception information in logs. But when I only set up one flink job(20 parallelism of source) and the cpu load of mysql server is less than 100%, that mistake never happens. mysql table: about 250 million data flink version: 1.18.1 mysql cdc connector version: 3.1.0 doris connector version :1.18-1.6.2 Is there anything to do to avoid this mistake if we can't promise the cpu load always in a healthy state? !image-2024-07-23-15-47-10-376.png! was: I use mysql cdc to read data from mysql to write to doris, mysql cdc may change from sanpshot mode to binlog mode failed when I set up 2 flink job(20 parallism of source) and the cpu load of mysql server is 100%. It can't change even the cpu load drop and just hang up, won't consume latest binlog data and there is no exception information in logs. But when I only set up one flink job(20 parallism of source) and the cpu load of mysql server is less than 100%, that mistake never happens. mysql table: about 250 million data flink version: 1.18.1 mysql cdc connector version: 3.1.0 doris connector version :1.18-1.6.2 Is there anything to do to avoid this mistake if we can't promise the cpu load always in a healthy state? !image-2024-07-23-15-47-10-376.png! > mysql cdc may change from snapshot mode to binlog mode failed when cpu load > of mysql server is 100% > --- > > Key: FLINK-35876 > URL: https://issues.apache.org/jira/browse/FLINK-35876 > Project: Flink > Issue Type: Bug > Components: Flink CDC >Affects Versions: cdc-3.1.0 >Reporter: Youle >Priority: Major > Attachments: image-2024-07-23-15-47-10-376.png > > > I use mysql cdc to read data from mysql to write to doris, mysql cdc may > change from sanpshot mode to binlog mode failed when I set up 2 flink job(20 > parallelism of source) and the cpu load of mysql server is 100%. It can't > change even the cpu load drop and just hang up, won't consume latest binlog > data and there is no exception information in logs. But when I only set up > one flink job(20 parallelism of source) and the cpu load of mysql server is > less than 100%, that mistake never happens. > mysql table: about 250 million data > flink version: 1.18.1 > mysql cdc connector version: 3.1.0 > doris connector version :1.18-1.6.2 > > Is there anything to do to avoid this mistake if we can't promise the cpu > load always in a healthy state? > !image-2024-07-23-15-47-10-376.png! -- This message was sent by Atlassian Jira (v8.20.10#820010)
[jira] [Created] (FLINK-36614) mysql cdc will be stuck when the ChangeEventQueue is full at snapshot consume phase
Youle created FLINK-36614: - Summary: mysql cdc will be stuck when the ChangeEventQueue is full at snapshot consume phase Key: FLINK-36614 URL: https://issues.apache.org/jira/browse/FLINK-36614 Project: Flink Issue Type: Bug Components: Flink CDC Affects Versions: cdc-3.1.1 Reporter: Youle Attachments: image-2024-10-28-20-21-11-112.png Reason: The ChangeEventQueue may get too many events because of little delay of network. The ChangeEventQueue is writen data to be full before it is called to disconnect the binarylog client connection. Solve: The data is not necessary when the the task is not running. I think we can avoid the ChangeEventQueue to be full by adding these code lines. !image-2024-10-28-20-21-11-112.png! Is it right to work in this way? Can anyone check this? -- This message was sent by Atlassian Jira (v8.20.10#820010)
