With two or more subscription URLs, start by grouping them by source and updating each group separately. Check the original server remarks before adding filters. Don’t delete servers just because they share a name—compare their connection settings and sources first. Finally, set separate update intervals for your primary and backup groups so an update doesn’t leave you searching for a configuration that was there before.
Subscription Groups: One Group per Source
A subscription URL retrieves a batch of configurations; a share link contains just one. Don’t paste a single share link into the subscription URL field, and don’t combine multiple sources in one group and sort them out later. A group name should tell you where its entries came from. Filter by region within a group rather than splitting groups by region.
- In the main window, open “Subscription Groups” → “Subscription Group Settings” and click “Add.” Enter an alias and subscription URL. For example, label the first URL “Source A | Primary” and the second “Source B | Backup.” There’s no need to include an account name, email address, or subscription token in an alias.
- After saving, select “Source A | Primary” in the main window and use the “Subscription Groups” menu to update the current subscription. Update this group by itself first, then repeat for the backup group. Don’t start with “Update All Subscriptions.” If the menu offers “through proxy” and “without proxy,” the former uses your existing proxy connection, while the latter requests the subscription URL directly. This setting only determines how the subscription is fetched this time.
- Leave the filter blank for the first update. For each group, note the number of entries, two or three original server remarks, and one configuration you’ve confirmed works. If the list suddenly becomes empty later, these notes can help you tell whether the subscription returned no results or the filter removed them all.
Server Filtering: Match Keywords in Remarks, Test One Group First
Use the main window’s search to find an entry in the current list. “Filter Regex” in a subscription group instead filters imported results by server remark when you update that group. After saving a filter, update the current group to check that it works; the existing list won’t show whether it took effect. A remark that says “Hong Kong” doesn’t prove the server is physically located there. Regex checks text only—it doesn’t test the connection or change routing.
Source A | Remarks use full region names
- Example original remarks
- Hong Kong 01, Japan 02, Singapore 03
- Filter regex
Hong Kong|Japan- Expected results
- Hong Kong 01, Japan 02
The vertical bar means either option can match. A match anywhere in the remark is enough.
Source B | Remarks start with region abbreviations
- Example original remarks
- hk-01、JP_02、SG-03
- Filter regex
(?i)^(HK|JP)[-_]- Expected results
- hk-01、JP_02
Case-insensitive; requires the remark to start with a region abbreviation followed by a specified separator.
This rule for the second group is more specific than searching for HK alone, and it will also miss Hong Kong-HK-01 and HK 01. Only require a prefix and separator if the provider consistently uses the same naming format. Don’t copy the Chinese filter from Group A into Group B: even when both offer the same region, their naming conventions may be completely different.
- Entries are still there, but the one you want is missing: Compare the original remarks to spot differences. For example, if the rule expects a hyphen but the provider switched to a space, loosen it to a broader keyword first. Once it matches, narrow it down again.
- The update reports success, but the group is empty: Clear the group’s filter regex and update the current group again. If entries return, revise the rule. If the group is still empty, check the update log, the subscription response, and whether the configurations can be parsed.
- Just looking for a server to test temporarily? Use the main window’s search without changing the subscription filter. That way, changing your search term won’t affect which entries are kept after the next update.
Duplicate or Unavailable Servers: Check Settings and Test Results, Not Just Names
Two sources listing “Hong Kong 01” doesn’t mean they point to the same server. Even the same address and port can use different protocols or authentication settings. Compare at least the protocol, address, port, authentication parameters, and transport. If the configuration uses TLS or other security settings, compare those parameters too. Treat two entries with different remarks as functional duplicates only when their connection settings match.
| What you see | What to check first | What to do |
|---|---|---|
| Two entries have the same remark | Source, protocol, and connection settings | If the settings differ, keep both and distinguish them by group. |
| A manually deleted entry comes back after an update | Check whether the entry is still in the source | If the naming pattern is consistent, filter within the group. If only a few configurations are duplicates, keep them and note their sources. |
| Only one entry in a group keeps failing | Under the same test conditions, can a known-working entry in the group connect? | If the comparison entry works, test the failing one again. If it continues to fail, disable that entry for now; there’s no need to delete the whole group. |
| Entries in both groups fail at the same time | Current network, core logs, and test target | First check for a general issue using a different working test target, then decide whether to remove any entries. |
To test availability, check the entries under investigation and a known-working entry on the same network and against the same test target. A latency result only means that test passed; it doesn’t guarantee that requests will work in practice. Also try a request from the application you normally use. Before deleting or filtering out your active configuration, select and verify another working one.
Deciding What to Remove: Find the Source of Duplicates First
If the same subscription URL was added twice, keep the group you’ve verified and remove the duplicate subscription entry. If different sources occasionally provide the same server, keep the source information rather than deleting an entire group just to remove one duplicate.
Update Intervals: Start with Daily for Primary and Weekly for Backup
Open “Subscription Groups” → “Subscription Group Settings,” then double-click a group and check the automatic update interval in the edit window. The values below are suggested starting points for organizing two subscription groups, not client defaults. If the field is in minutes, enter the converted values shown. If your version uses a different unit, convert according to the unit shown in the interface. Make sure automatic updates are enabled for both the group and the relevant client settings.
- Primary group: If the provider has just changed its configurations, update the current group right away instead of waiting until tomorrow. Before updating, note the remark for the configuration you’re using. Afterward, confirm it’s still there and make a real request.
- Backup group: You can update less often if its configurations rarely change, but don’t wait until your primary connection fails to test it for the first time. If the backup subscription can only be fetched through your current proxy, update the backup group and confirm it has a working entry while the primary connection is still available.
- Automatic updates aren’t running: First make sure the client is running during the expected update window. Then check the update log to see whether the update wasn’t triggered or the fetch failed. A successful manual update only confirms the subscription could be fetched at that moment; it doesn’t mean the schedule is working.
Update Frequency Can’t Fix a Filter Rule
If a provider’s remark changes and the filter drops the same servers every time, loosen the group’s rule first and update that group alone to check the new remark format. Then restore your usual schedule. Shortening the update interval will only make the same result happen more often.