File names over 230 characters will cause JGit exception during source control operationsDescriptionIf a file name exists that is over 230 characters, and therefore close to the maximum Unix filename length of 255 characters, then source control operations can fail. Inspecting the logs show the following exception: JGitInternalException: Exception caught during execution of reset command. File name too long This is because jGit generates temporary files using the prefix "._" and then the update's name, followed by a random long number (maximum of 19 digits) and a final suffix of ".tmp". Hence 255 - 2 - 4 - 19 = 230 characters is the limit and anything greater than this will generate a file name longer than is supported by the underlying operating system. Given a random long value is used as part of the filename, it is possible that the issue is not seen every time, but is more likely the longer the update name is over 230 characters.Steps to Reproduce One easy way to get a long file name is to create a compound index with a lot of columns, as the generated name will contain each column on the index. The resulting file name will follow this general format: sys_index_table_name_column_a_column_b_column_c_column_d_column_e.xmlWorkaroundAfter carefully considering the severity and frequency of the issue, and the cost and risk of attempting a fix, it has been decided to not address this issue in any current or near future releases. We do not make this decision lightly, and we apologise for any inconvenience. The workaround is to avoid creating compound indexes with more than a small amount of columns. This is a preventative measure. In case the issue has already occurred, you will no longer be able to perform source control operations with the existing repository. The solution in that case is to delete or reduce the complexity of the compound index, unlink your application from source control, and then re-link it to a brand new empty repository.Related Problem: PRB1307874