I am doing something similar for 2005-2013, and I don’t have missing values of the household identifiers, but I did have to obtain the March identifiers from the NBER.
The HRHHID and HRHHID2 identifiers are only available for the non-March months (check the availability of HRHHID and HRHHID2). Did you happen to check whether the observations you have missing are from the March months?
If indeed the missing IDs are from the March files, you can download the NBER CPS March supplements, eliminate non-interview individuals (which are not shown in the IPUMS-CPS) and sequentially merge the March files of the NBER with the IPUMS files. When you sequentially merge you can also opt to keep only the ID variables from the NBER (since you can obtain the others in a nicer, harmonized format from the IPUMS)