The Refresh Service task and Clear REST Cache operation call in ArcGIS REST API do not update the time extent information on multidimensional image services on all machines participating in a High Availability (HA) environment on an intermittent basis.
上次发布: January 6, 2023ArcGIS GIS Server
漏洞 ID 编号
BUG-000153780
已提交
November 17, 2022
上次修改时间
June 5, 2024
适用范围
ArcGIS GIS Server
找到的版本
10.9.1
操作系统
N/A
操作系统版本
N/A
状态
Known Limit
经开发团队审核,已确定此问题与不受 Esri 控制的软件的已知限制有关。 问题的“其他信息”部分可能包含进一步说明。
附加信息
This is a known issue. It is because it has multiple server nodes with multiple instances. Some instances are refreshed, while others are not.
When you send a refresh service call, the corresponding image service is not refreshed automatically. Instead, it generates a flag for the image service. The next call to the image service destroys the existing instance and creates a new one to pick up new data in the mosaic dataset.
However, this might conflict with another server functionality. By default, the server system cleans up the folder that stores the refresh flag file when it reaches 10 minutes old. So, if the new image service call is received after 10 minutes of the image service refresh, the refresh is not going to take effect. This does not mean the refresh service call has failed, as noticed. It is just the flag that is deleted without affecting anything.
解决办法
A few settings to try:
Make the 'keep alive interval' property of the image service less than the ArcGIS output folder 'max file age' (https://server/arcgis/admin/system/directories/arcgisoutput). In this way, bad instances are killed before the flag file (which is stored in the ArcGIS output folder) is deleted.
Send some dummy image server calls at the same time right after the image service refresh flag is generated. This is to make sure all instances are hit and refreshed.